20200915.md

ArXiv cs.CV --Tue, 15 Sep 2020

1.High-Resolution Deep Image Matting ⬇️

Image matting is a key technique for image and video editing and composition. Conventionally, deep learning approaches take the whole input image and an associated trimap to infer the alpha matte using convolutional neural networks. Such approaches set state-of-the-arts in image matting; however, they may fail in real-world matting applications due to hardware limitations, since real-world input images for matting are mostly of very high resolution. In this paper, we propose HDMatt, a first deep learning based image matting approach for high-resolution inputs. More concretely, HDMatt runs matting in a patch-based crop-and-stitch manner for high-resolution inputs with a novel module design to address the contextual dependency and consistency issues between different patches. Compared with vanilla patch-based inference which computes each patch independently, we explicitly model the cross-patch contextual dependency with a newly-proposed Cross-Patch Contextual module (CPC) guided by the given trimap. Extensive experiments demonstrate the effectiveness of the proposed method and its necessity for high-resolution inputs. Our HDMatt approach also sets new state-of-the-art performance on Adobe Image Matting and AlphaMatting benchmarks and produce impressive visual results on more real-world high-resolution images.

2.Adaptive Text Recognition through Visual Matching ⬇️

In this work, our objective is to address the problems of generalization and flexibility for text recognition in documents. We introduce a new model that exploits the repetitive nature of characters in languages, and decouples the visual representation learning and linguistic modelling stages. By doing this, we turn text recognition into a shape matching problem, and thereby achieve generalization in appearance and flexibility in classes. We evaluate the new model on both synthetic and real datasets across different alphabets and show that it can handle challenges that traditional architectures are not able to solve without expensive retraining, including: (i) it can generalize to unseen fonts without new exemplars from them; (ii) it can flexibly change the number of classes, simply by changing the exemplars provided; and (iii) it can generalize to new languages and new characters that it has not been trained for by providing a new glyph set. We show significant improvements over state-of-the-art models for all these cases.

3.GIA-Net: Global Information Aware Network for Low-light Imaging ⬇️

It is extremely challenging to acquire perceptually plausible images under low-light conditions due to low SNR. Most recently, U-Nets have shown promising results for low-light imaging. However, vanilla U-Nets generate images with artifacts such as color inconsistency due to the lack of global color information. In this paper, we propose a global information aware (GIA) module, which is capable of extracting and integrating the global information into the network to improve the performance of low-light imaging. The GIA module can be inserted into a vanilla U-Net with negligible extra learnable parameters or computational cost. Moreover, a GIA-Net is constructed, trained and evaluated on a large scale real-world low-light imaging dataset. Experimental results show that the proposed GIA-Net outperforms the state-of-the-art methods in terms of four metrics, including deep metrics that measure perceptual similarities. Extensive ablation studies have been conducted to verify the effectiveness of the proposed GIA-Net for low-light imaging by utilizing global information.

4.Collaborative Attention Mechanism for Multi-View Action Recognition ⬇️

Multi-view action recognition (MVAR) leverages complementary temporal information from different views to enhance the learning process. Attention is an effective mechanism which has been extensively adopted for modeling temporal data. However, most existing MVAR methods only utilize attention to extract view-specific patterns. They ignore the potential to dig latent mutual-support information inattention space. To fully take the advantage of the multi-view cooperation, we propose a collaborative attention mechanism (CAM). It detects the attention differences among multi-view inputs, and adaptively integrates complementary frame-level information to benefit each other. Specifically, we utilize recurrent neural network (RNN) by expanding long short-term memory (LSTM) as a Mutual-Aid RNN (MAR). CAM takes advantages of view-specific attention pattern to guide another view and unlock potential information which is hard to explore by itself. Extensive experiments on three action datasets illustrate our CAM achieves better result for each single view, and also boosts the multi-view performance.

5.Zero-shot Synthesis with Group-Supervised Learning ⬇️

Visual cognition of primates is superior to that of artificial neural networks in its ability to 'envision' a visual object, even a newly-introduced one, in different attributes including pose, position, color, texture, etc. To aid neural networks to envision objects with different attributes, we propose a family of objective functions, expressed on groups of examples, as a novel learning framework that we term Group-Supervised Learning (GSL). GSL decomposes inputs into a disentangled representation with swappable components that can be recombined to synthesize new samples, trained through similarity mining within groups of exemplars. For instance, images of red boats & blue cars can be decomposed and recombined to synthesize novel images of red cars. We describe a general class of datasets admissible by GSL. We propose an implementation based on auto-encoder, termed group-supervised zero-shot synthesis network (GZS-Net) trained with our learning framework, that can produce a high-quality red car even if no such example is witnessed during training. We test our model and learning framework on existing benchmarks, in addition to new dataset that we open-source. We qualitatively and quantitatively demonstrate that GZS-Net trained with GSL outperforms state-of-the-art methods

6.Beyond Weak Perspective for Monocular 3D Human Pose Estimation ⬇️

We consider the task of 3D joints location and orientation prediction from a monocular video with the skinned multi-person linear (SMPL) model. We first infer 2D joints locations with an off-the-shelf pose estimation algorithm. We use the SPIN algorithm and estimate initial predictions of body pose, shape and camera parameters from a deep regression neural network. We then adhere to the SMPLify algorithm which receives those initial parameters, and optimizes them so that inferred 3D joints from the SMPL model would fit the 2D joints locations. This algorithm involves a projection step of 3D joints to the 2D image plane. The conventional approach is to follow weak perspective assumptions which use ad-hoc focal length. Through experimentation on the 3D Poses in the Wild (3DPW) dataset, we show that using full perspective projection, with the correct camera center and an approximated focal length, provides favorable results. Our algorithm has resulted in a winning entry for the 3DPW Challenge, reaching first place in joints orientation accuracy.

7.Improving Inversion and Generation Diversity in StyleGAN using a Gaussianized Latent Space ⬇️

Modern Generative Adversarial Networks are capable of creating artificial, photorealistic images from latent vectors living in a low-dimensional learned latent space. It has been shown that a wide range of images can be projected into this space, including images outside of the domain that the generator was trained on. However, while in this case the generator reproduces the pixels and textures of the images, the reconstructed latent vectors are unstable and small perturbations result in significant image distortions. In this work, we propose to explicitly model the data distribution in latent space. We show that, under a simple nonlinear operation, the data distribution can be modeled as Gaussian and therefore expressed using sufficient statistics. This yields a simple Gaussian prior, which we use to regularize the projection of images into the latent space. The resulting projections lie in smoother and better behaved regions of the latent space, as shown using interpolation performance for both real and generated images. Furthermore, the Gaussian model of the distribution in latent space allows us to investigate the origins of artifacts in the generator output, and provides a method for reducing these artifacts while maintaining diversity of the generated images.

8.A Study of Human Gaze Behavior During Visual Crowd Counting ⬇️

In this paper, we describe our study on how humans allocate their attention during visual crowd counting. Using an eye tracker, we collect gaze behavior of human participants who are tasked with counting the number of people in crowd images. Analyzing the collected gaze behavior of ten human participants on thirty crowd images, we observe some common approaches for visual counting. For an image of a small crowd, the approach is to enumerate over all people or groups of people in the crowd, and this explains the high level of similarity between the fixation density maps of different human participants. For an image of a large crowd, our participants tend to focus on one section of the image, count the number of people in that section, and then extrapolate to the other sections. In terms of count accuracy, our human participants are not as good at the counting task, compared to the performance of the current state-of-the-art computer algorithms. Interestingly, there is a tendency to under count the number of people in all crowd images. Gaze behavior data and images can be downloaded from this https URL

9.Fast Implementation of 4-bit Convolutional Neural Networks for Mobile Devices ⬇️

Quantized low-precision neural networks are very popular because they require less computational resources for inference and can provide high performance, which is vital for real-time and embedded recognition systems. However, their advantages are apparent for FPGA and ASIC devices, while general-purpose processor architectures are not always able to perform low-bit integer computations efficiently. The most frequently used low-precision neural network model for mobile central processors is an 8-bit quantized network. However, in a number of cases, it is possible to use fewer bits for weights and activations, and the only problem is the difficulty of efficient implementation. We introduce an efficient implementation of 4-bit matrix multiplication for quantized neural networks and perform time measurements on a mobile ARM processor. It shows 2.9 times speedup compared to standard floating-point multiplication and is 1.5 times faster than 8-bit quantized one. We also demonstrate a 4-bit quantized neural network for OCR recognition on the MIDV-500 dataset. 4-bit quantization gives 95.0% accuracy and 48% overall inference speedup, while an 8-bit quantized network gives 95.4% accuracy and 39% speedup. The results show that 4-bit quantization perfectly suits mobile devices, yielding good enough accuracy and low inference time.

10.Unsupervised Domain Adaptation by Uncertain Feature Alignment ⬇️

Unsupervised domain adaptation (UDA) deals with the adaptation of models from a given source domain with labeled data to an unlabeled target domain. In this paper, we utilize the inherent prediction uncertainty of a model to accomplish the domain adaptation task. The uncertainty is measured by Monte-Carlo dropout and used for our proposed Uncertainty-based Filtering and Feature Alignment (UFAL) that combines an Uncertain Feature Loss (UFL) function and an Uncertainty-Based Filtering (UBF) approach for alignment of features in Euclidean space. Our method surpasses recently proposed architectures and achieves state-of-the-art results on multiple challenging datasets. Code is available on the project website.

11.EfficientSeg: An Efficient Semantic Segmentation Network ⬇️

Deep neural network training without pre-trained weights and few data is shown to need more training iterations. It is also known that, deeper models are more successful than their shallow counterparts for semantic segmentation task. Thus, we introduce EfficientSeg architecture, a modified and scalable version of U-Net, which can be efficiently trained despite its depth. We evaluated EfficientSeg architecture on Minicity dataset and outperformed U-Net baseline score (40% mIoU) using the same parameter count (51.5% mIoU). Our most successful model obtained 58.1% mIoU score and got the fourth place in semantic segmentation track of ECCV 2020 VIPriors challenge.

12.Scene-Graph Augmented Data-Driven Risk Assessment of Autonomous Vehicle Decisions ⬇️

Despite impressive advancements in Autonomous Driving Systems (ADS), navigation in complex road conditions remains a challenging problem. There is considerable evidence that evaluating the subjective risk level of various decisions can improve ADS' safety in both normal and complex driving scenarios. However, existing deep learning-based methods often fail to model the relationships between traffic participants and can suffer when faced with complex real-world scenarios. Besides, these methods lack transferability and explainability. To address these limitations, we propose a novel data-driven approach that uses scene-graphs as intermediate representations. Our approach includes a Multi-Relation Graph Convolution Network, a Long-Short Term Memory Network, and attention layers for modeling the subjective risk of driving maneuvers. To train our model, we formulate this task as a supervised scene classification problem. We consider a typical use case to demonstrate our model's capabilities: lane changes. We show that our approach achieves a higher classification accuracy than the state-of-the-art approach on both large (96.4% vs. 91.2%) and small (91.8% vs. 71.2%) synthesized datasets, also illustrating that our approach can learn effectively even from smaller datasets. We also show that our model trained on a synthesized dataset achieves an average accuracy of 87.8% when tested on a real-world dataset compared to the 70.3% accuracy achieved by the state-of-the-art model trained on the same synthesized dataset, showing that our approach can more effectively transfer knowledge. Finally, we demonstrate that the use of spatial and temporal attention layers improves our model's performance by 2.7% and 0.7% respectively, and increases its explainability.

13.Adaptive Label Smoothing ⬇️

This paper concerns the use of objectness measures to improve the calibration performance of Convolutional Neural Networks (CNNs). Objectness is a measure of likelihood of an object from any class being present in a given image. CNNs have proven to be very good classifiers and generally localize objects well; however, the loss functions typically used to train classification CNNs do not penalize inability to localize an object, nor do they take into account an object's relative size in the given image. We present a novel approach to object localization that combines the ideas of objectness and label smoothing during training. Unlike previous methods, we compute a smoothing factor that is adaptive based on relative object size within an image. We present extensive results using ImageNet and OpenImages to demonstrate that CNNs trained using adaptive label smoothing are much less likely to be overconfident in their predictions, as compared to CNNs trained using hard targets. We also show qualitative results using class activation maps to illustrate the improvements.

14.Completely Self-Supervised Crowd Counting via Distribution Matching ⬇️

Dense crowd counting is a challenging task that demands millions of head annotations for training models. Though existing self-supervised approaches could learn good representations, they require some labeled data to map these features to the end task of density estimation. We mitigate this issue with the proposed paradigm of complete self-supervision, which does not need even a single labeled image. The only input required to train, apart from a large set of unlabeled crowd images, is the approximate upper limit of the crowd count for the given dataset. Our method dwells on the idea that natural crowds follow a power law distribution, which could be leveraged to yield error signals for backpropagation. A density regressor is first pretrained with self-supervision and then the distribution of predictions is matched to the prior by optimizing Sinkhorn distance between the two. Experiments show that this results in effective learning of crowd features and delivers significant counting performance. Furthermore, we establish the superiority of our method in less data setting as well. The code and models for our approach is available at this https URL.

15.Synbols: Probing Learning Algorithms with Synthetic Datasets ⬇️

Progress in the field of machine learning has been fueled by the introduction of benchmark datasets pushing the limits of existing algorithms. Enabling the design of datasets to test specific properties and failure modes of learning algorithms is thus a problem of high interest, as it has a direct impact on innovation in the field. In this sense, we introduce Synbols -- Synthetic Symbols -- a tool for rapidly generating new datasets with a rich composition of latent features rendered in low resolution images. Synbols leverages the large amount of symbols available in the Unicode standard and the wide range of artistic font provided by the open font community. Our tool's high-level interface provides a language for rapidly generating new distributions on the latent features, including various types of textures and occlusions. To showcase the versatility of Synbols, we use it to dissect the limitations and flaws in standard learning algorithms in various learning setups including supervised learning, active learning, out of distribution generalization, unsupervised representation learning, and object counting.

16.Adaptive Convolution Kernel for Artificial Neural Networks ⬇️

Many deep neural networks are built by using stacked convolutional layers of fixed and single size (often 3$\times$3) kernels. This paper describes a method for training the size of convolutional kernels to provide varying size kernels in a single layer. The method utilizes a differentiable, and therefore backpropagation-trainable Gaussian envelope which can grow or shrink in a base grid. Our experiments compared the proposed adaptive layers to ordinary convolution layers in a simple two-layer network, a deeper residual network, and a U-Net architecture. The results in the popular image classification datasets such as MNIST, MNIST-CLUTTERED, CIFAR-10, Fashion, and ``Faces in the Wild'' showed that the adaptive kernels can provide statistically significant improvements on ordinary convolution kernels. A segmentation experiment in the Oxford-Pets dataset demonstrated that replacing a single ordinary convolution layer in a U-shaped network with a single 7$\times$7 adaptive layer can improve its learning performance and ability to generalize.

17.P-DIFF: Learning Classifier with Noisy Labels based on Probability Difference Distributions ⬇️

Learning deep neural network (DNN) classifier with noisy labels is a challenging task because the DNN can easily over-fit on these noisy labels due to its high capability. In this paper, we present a very simple but effective training paradigm called P-DIFF, which can train DNN classifiers but obviously alleviate the adverse impact of noisy labels. Our proposed probability difference distribution implicitly reflects the probability of a training sample to be clean, then this probability is employed to re-weight the corresponding sample during the training process. P-DIFF can also achieve good performance even without prior knowledge on the noise rate of training samples. Experiments on benchmark datasets also demonstrate that P-DIFF is superior to the state-of-the-art sample selection methods.

18.4Seasons: A Cross-Season Dataset for Multi-Weather SLAM in Autonomous Driving ⬇️

We present a novel dataset covering seasonal and challenging perceptual conditions for autonomous driving. Among others, it enables research on visual odometry, global place recognition, and map-based re-localization tracking. The data was collected in different scenarios and under a wide variety of weather conditions and illuminations, including day and night. This resulted in more than 350 km of recordings in nine different environments ranging from multi-level parking garage over urban (including tunnels) to countryside and highway. We provide globally consistent reference poses with up-to centimeter accuracy obtained from the fusion of direct stereo visual-inertial odometry with RTK-GNSS. The full dataset is available at this http URL.

19.PRAFlow_RVC: Pyramid Recurrent All-Pairs Field Transforms for Optical Flow Estimation in Robust Vision Challenge 2020 ⬇️

Optical flow estimation is an important computer vision task, which aims at estimating the dense correspondences between two frames. RAFT (Recurrent All Pairs Field Transforms) currently represents the state-of-the-art in optical flow estimation. It has excellent generalization ability and has obtained outstanding results across several benchmarks. To further improve the robustness and achieve accurate optical flow estimation, we present PRAFlow (Pyramid Recurrent All-Pairs Flow), which builds upon the pyramid network structure. Due to computational limitation, our proposed network structure only uses two pyramid layers. At each layer, the RAFT unit is used to estimate the optical flow at the current resolution. Our model was trained on several simulate and real-image datasets, submitted to multiple leaderboards using the same model and parameters, and won the 2nd place in the optical flow task of ECCV 2020 workshop: Robust Vision Challenge.

20.DeepWriteSYN: On-Line Handwriting Synthesis via Deep Short-Term Representations ⬇️

This study proposes DeepWriteSYN, a novel on-line handwriting synthesis approach via deep short-term representations. It comprises two modules: i) an optional and interchangeable temporal segmentation, which divides the handwriting into short-time segments consisting of individual or multiple concatenated strokes; and ii) the on-line synthesis of those short-time handwriting segments, which is based on a sequence-to-sequence Variational Autoencoder (VAE). The main advantages of the proposed approach are that the synthesis is carried out in short-time segments (that can run from a character fraction to full characters) and that the VAE can be trained on a configurable handwriting dataset. These two properties give a lot of flexibility to our synthesiser, e.g., as shown in our experiments, DeepWriteSYN can generate realistic handwriting variations of a given handwritten structure corresponding to the natural variation within a given population or a given subject. These two cases are developed experimentally for individual digits and handwriting signatures, respectively, achieving in both cases remarkable results.
Also, we provide experimental results for the task of on-line signature verification showing the high potential of DeepWriteSYN to improve significantly one-shot learning scenarios. To the best of our knowledge, this is the first synthesis approach capable of generating realistic on-line handwriting in the short term (including handwritten signatures) via deep learning. This can be very useful as a module toward long-term realistic handwriting generation either completely synthetic or as natural variation of given handwriting samples.

21.Deep intrinsic decomposition trained on surreal scenes yet with realistic light effects ⬇️

Estimation of intrinsic images still remains a challenging task due to weaknesses of ground-truth datasets, which either are too small or present non-realistic issues. On the other hand, end-to-end deep learning architectures start to achieve interesting results that we believe could be improved if important physical hints were not ignored. In this work, we present a twofold framework: (a) a flexible generation of images overcoming some classical dataset problems such as larger size jointly with coherent lighting appearance; and (b) a flexible architecture tying physical properties through intrinsic losses. Our proposal is versatile, presents low computation time, and achieves state-of-the-art results.

22.AIM 2020 Challenge on Video Extreme Super-Resolution: Methods and Results ⬇️

This paper reviews the video extreme super-resolution challenge associated with the AIM 2020 workshop at ECCV 2020. Common scaling factors for learned video super-resolution (VSR) do not go beyond factor 4. Missing information can be restored well in this region, especially in HR videos, where the high-frequency content mostly consists of texture details. The task in this challenge is to upscale videos with an extreme factor of 16, which results in more serious degradations that also affect the structural integrity of the videos. A single pixel in the low-resolution (LR) domain corresponds to 256 pixels in the high-resolution (HR) domain. Due to this massive information loss, it is hard to accurately restore the missing information. Track 1 is set up to gauge the state-of-the-art for such a demanding task, where fidelity to the ground truth is measured by PSNR and SSIM. Perceptually higher quality can be achieved in trade-off for fidelity by generating plausible high-frequency content. Track 2 therefore aims at generating visually pleasing results, which are ranked according to human perception, evaluated by a user study. In contrast to single image super-resolution (SISR), VSR can benefit from additional information in the temporal domain. However, this also imposes an additional requirement, as the generated frames need to be consistent along time.

23.Unsupervised learning for vascular heterogeneity assessment of glioblastoma based on magnetic resonance imaging: The Hemodynamic Tissue Signature ⬇️

This thesis focuses on the research and development of the Hemodynamic Tissue Signature (HTS) method: an unsupervised machine learning approach to describe the vascular heterogeneity of glioblastomas by means of perfusion MRI analysis. The HTS builds on the concept of habitats. An habitat is defined as a sub-region of the lesion with a particular MRI profile describing a specific physiological behavior. The HTS method delineates four habitats within the glioblastoma: the High Angiogenic Tumor (HAT) habitat, as the most perfused region of the enhancing tumor; the Low Angiogenic Tumor (LAT) habitat, as the region of the enhancing tumor with a lower angiogenic profile; the potentially Infiltrated Peripheral Edema (IPE) habitat, as the non-enhancing region adjacent to the tumor with elevated perfusion indexes; and the Vasogenic Peripheral Edema (VPE) habitat, as the remaining edema of the lesion with the lowest perfusion profile.
The results of this thesis have been published in ten scientific contributions, including top-ranked journals and conferences in the areas of Medical Informatics, Statistics and Probability, Radiology & Nuclear Medicine, Machine Learning and Data Mining and Biomedical Engineering. An industrial patent registered in Spain (ES201431289A), Europe (EP3190542A1) and EEUU (US20170287133A1) was also issued, summarizing the efforts of the thesis to generate tangible assets besides the academic revenue obtained from research publications. Finally, the methods, technologies and original ideas conceived in this thesis led to the foundation of ONCOANALYTICS CDX, a company framed into the business model of companion diagnostics for pharmaceutical compounds, thought as a vehicle to facilitate the industrialization of the ONCOhabitats technology.

24.Accurate and Lightweight Image Super-Resolution with Model-Guided Deep Unfolding Network ⬇️

Deep neural networks (DNNs) based methods have achieved great success in single image super-resolution (SISR). However, existing state-of-the-art SISR techniques are designed like black boxes lacking transparency and interpretability. Moreover, the improvement in visual quality is often at the price of increased model complexity due to black-box design. In this paper, we present and advocate an explainable approach toward SISR named model-guided deep unfolding network (MoG-DUN). Targeting at breaking the coherence barrier, we opt to work with a well-established image prior named nonlocal auto-regressive model and use it to guide our DNN design. By integrating deep denoising and nonlocal regularization as trainable modules within a deep learning framework, we can unfold the iterative process of model-based SISR into a multi-stage concatenation of building blocks with three interconnected modules (denoising, nonlocal-AR, and reconstruction). The design of all three modules leverages the latest advances including dense/skip connections as well as fast nonlocal implementation. In addition to explainability, MoG-DUN is accurate (producing fewer aliasing artifacts), computationally efficient (with reduced model parameters), and versatile (capable of handling multiple degradations). The superiority of the proposed MoG-DUN method to existing state-of-the-art image SR methods including RCAN, SRMDNF, and SRFBN is substantiated by extensive experiments on several popular datasets and various degradation scenarios.

25.Prior Knowledge about Attributes: Learning a More Effective Potential Space for Zero-Shot Recognition ⬇️

Zero-shot learning (ZSL) aims to recognize unseen classes accurately by learning seen classes and known attributes, but correlations in attributes were ignored by previous study which lead to classification results confused. To solve this problem, we build an Attribute Correlation Potential Space Generation (ACPSG) model which uses a graph convolution network and attribute correlation to generate a more discriminating potential space. Combining potential discrimination space and user-defined attribute space, we can better classify unseen classes. Our approach outperforms some existing state-of-the-art methods on several benchmark datasets, whether it is conventional ZSL or generalized ZSL.

26.Cascade Network for Self-Supervised Monocular Depth Estimation ⬇️

It is a classical compute vision problem to obtain real scene depth maps by using a monocular camera, which has been widely concerned in recent years. However, training this model usually requires a large number of artificially labeled samples. To solve this problem, some researchers use a self-supervised learning model to overcome this problem and reduce the dependence on manually labeled data. Nevertheless, the accuracy and reliability of these methods have not reached the expected standard. In this paper, we propose a new self-supervised learning method based on cascade networks. Compared with the previous self-supervised methods, our method has improved accuracy and reliability, and we have proved this by experiments. We show a cascaded neural network that divides the target scene into parts of different sight distances and trains them separately to generate a better depth map. Our approach is divided into the following four steps. In the first step, we use the self-supervised model to estimate the depth of the scene roughly. In the second step, the depth of the scene generated in the first step is used as a label to divide the scene into different depth parts. The third step is to use models with different parameters to generate depth maps of different depth parts in the target scene, and the fourth step is to fuse the depth map. Through the ablation study, we demonstrated the effectiveness of each component individually and showed high-quality, state-of-the-art results in the KITTI benchmark.

27.Residual Learning for Effective joint Demosaicing-Denoising ⬇️

Image demosaicing and denoising are key steps for color image production pipeline. The classical processing sequence consists in applying denoising first, and then demosaicing. However, this sequence leads to oversmoothing and unpleasant checkerboard effect. Moreover, it is very difficult to change this order, because once the image is demosaiced, the statistical properties of the noise will be changed dramatically. This is extremely challenging for the traditional denoising models that strongly rely on statistical assumptions. In this paper, we attempt to tackle this prickly problem. Indeed, here we invert the traditional CFA processing pipeline by first applying demosaicing and then using an adapted denoising. In order to obtain high-quality demosaicing of noiseless images, we combine the advantages of traditional algorithms with deep learning. This is achieved by training convolutional neural networks (CNNs) to learn the residuals of traditional algorithms. To improve the performance in image demosaicing we propose a modified Inception architecture. Given the trained demosaicing as a basic component, we apply it to noisy images and use another CNN to learn the residual noise (including artifacts) of the demosaiced images, which allows to reconstruct full color images. Experimental results show clearly that this method outperforms several state-of-the-art methods both quantitatively as well as in terms of visual quality.

28.Learning from Multimodal and Multitemporal Earth Observation Data for Building Damage Mapping ⬇️

Earth observation technologies, such as optical imaging and synthetic aperture radar (SAR), provide excellent means to monitor ever-growing urban environments continuously. Notably, in the case of large-scale disasters (e.g., tsunamis and earthquakes), in which a response is highly time-critical, images from both data modalities can complement each other to accurately convey the full damage condition in the disaster's aftermath. However, due to several factors, such as weather and satellite coverage, it is often uncertain which data modality will be the first available for rapid disaster response efforts. Hence, novel methodologies that can utilize all accessible EO datasets are essential for disaster management. In this study, we have developed a global multisensor and multitemporal dataset for building damage mapping. We included building damage characteristics from three disaster types, namely, earthquakes, tsunamis, and typhoons, and considered three building damage categories. The global dataset contains high-resolution optical imagery and high-to-moderate-resolution multiband SAR data acquired before and after each disaster. Using this comprehensive dataset, we analyzed five data modality scenarios for damage mapping: single-mode (optical and SAR datasets), cross-modal (pre-disaster optical and post-disaster SAR datasets), and mode fusion scenarios. We defined a damage mapping framework for the semantic segmentation of damaged buildings based on a deep convolutional neural network algorithm. We compare our approach to another state-of-the-art baseline model for damage mapping. The results indicated that our dataset, together with a deep learning network, enabled acceptable predictions for all the data modality scenarios.

29.RelativeNAS: Relative Neural Architecture Search via Slow-Fast Learning ⬇️

Despite the remarkable successes of Convolutional neural networks (CNN) in computer vision, it is time-consuming and error-prone to manually design a CNN. Among various neural architecture search (NAS) methods that are motivated to automate designs of high-performance CNNs, the differentiable NAS and population-based NAS are attracting increasing interests due to their unique characters. To benefit from the merits while overcoming the deficiencies of both, this work proposes a novel NAS method, RelativeNAS. As the key to efficient search, RelativeNAS performs joint learning between fast-learners (i.e. networks with relatively higher accuracy) and slow-learners in a pairwise manner. Moreover, since RelativeNAS only requires low-fidelity performance estimation to distinguish each pair of fast-learner and slow-learner, it saves certain computation costs for training the candidate architectures. The proposed RelativeNAS brings several unique advantages: (1) it achieves state-of-the-art performances on imageNet with top-1 error rate of 24.88%, i.e. outperforming DARTS and AmoebaNet-B by 1.82% and 1.12% respectively; (2) it spends only nine hours with a single 1080Ti GPU to obtain the discovered cells, i.e. 3.75x and 7875x faster than DARTS and AmoebaNet respectively; (3) it provides that the discovered cells obtained on CIFAR-10 can be directly transferred to object detection, semantic segmentation, and keypoint detection, yielding competitive results of 73.1% mAP on PASCAL VOC, 78.7% mIoU on Cityscapes, and 68.5% AP on MSCOCO, respectively. The code is available at this https URL

30.3D Object Detection and Tracking Based on Streaming Data ⬇️

Recent approaches for 3D object detection have made tremendous progresses due to the development of deep learning. However, previous researches are mostly based on individual frames, leading to limited exploitation of information between frames. In this paper, we attempt to leverage the temporal information in streaming data and explore 3D streaming based object detection as well as tracking. Toward this goal, we set up a dual-way network for 3D object detection based on keyframes, and then propagate predictions to non-key frames through a motion based interpolation algorithm guided by temporal information. Our framework is not only shown to have significant improvements on object detection compared with frame-by-frame paradigm, but also proven to produce competitive results on KITTI Object Tracking Benchmark, with 76.68% in MOTA and 81.65% in MOTP respectively.

31.One-bit Supervision for Image Classification ⬇️

This paper presents one-bit supervision, a novel setting of learning from incomplete annotations, in the scenario of image classification. Instead of training a model upon the accurate label of each sample, our setting requires the model to query with a predicted label of each sample and learn from the answer whether the guess is correct. This provides one bit (yes or no) of information, and more importantly, annotating each sample becomes much easier than finding the accurate label from many candidate classes. There are two keys to training a model upon one-bit supervision: improving the guess accuracy and making use of incorrect guesses. For these purposes, we propose a multi-stage training paradigm which incorporates negative label suppression into an off-the-shelf semi-supervised learning algorithm. In three popular image classification benchmarks, our approach claims higher efficiency in utilizing the limited amount of annotations.

32.GINet: Graph Interaction Network for Scene Parsing ⬇️

Recently, context reasoning using image regions beyond local convolution has shown great potential for scene parsing. In this work, we explore how to incorporate the linguistic knowledge to promote context reasoning over image regions by proposing a Graph Interaction unit (GI unit) and a Semantic Context Loss (SC-loss). The GI unit is capable of enhancing feature representations of convolution networks over high-level semantics and learning the semantic coherency adaptively to each sample. Specifically, the dataset-based linguistic knowledge is first incorporated in the GI unit to promote context reasoning over the visual graph, then the evolved representations of the visual graph are mapped to each local representation to enhance the discriminated capability for scene parsing. GI unit is further improved by the SC-loss to enhance the semantic representations over the exemplar-based semantic graph. We perform full ablation studies to demonstrate the effectiveness of each component in our approach. Particularly, the proposed GINet outperforms the state-of-the-art approaches on the popular benchmarks, including Pascal-Context and COCO Stuff.

33.SCOUTER: Slot Attention-based Classifier for Explainable Image Recognition ⬇️

Explainable artificial intelligence is gaining attention. However, most existing methods are based on gradients or intermediate features, which are not directly involved in the decision-making process of the classifier. In this paper, we propose a slot attention-based light-weighted classifier called SCOUTER for transparent yet accurate classification. Two major differences from other attention-based methods include: (a) SCOUTER's explanation involves the final confidence for each category, offering more intuitive interpretation, and (b) all the categories have their corresponding positive or negative explanation, which tells "why the image is of a certain category" or "why the image is not of a certain category." We design a new loss tailored for SCOUTER that controls the model's behavior to switch between positive and negative explanations, as well as the size of explanatory regions. Experimental results show that SCOUTER can give better visual explanations while keeping good accuracy on a large dataset.

34.Accelerating COVID-19 Differential Diagnosis with Explainable Ultrasound Image Analysis ⬇️

Controlling the COVID-19 pandemic largely hinges upon the existence of fast, safe, and highly-available diagnostic tools. Ultrasound, in contrast to CT or X-Ray, has many practical advantages and can serve as a globally-applicable first-line examination technique. We provide the largest publicly available lung ultrasound (US) dataset for COVID-19 consisting of 106 videos from three classes (COVID-19, bacterial pneumonia, and healthy controls); curated and approved by medical experts. On this dataset, we perform an in-depth study of the value of deep learning methods for differential diagnosis of COVID-19. We propose a frame-based convolutional neural network that correctly classifies COVID-19 US videos with a sensitivity of 0.98+-0.04 and a specificity of 0.91+-08 (frame-based sensitivity 0.93+-0.05, specificity 0.87+-0.07). We further employ class activation maps for the spatio-temporal localization of pulmonary biomarkers, which we subsequently validate for human-in-the-loop scenarios in a blindfolded study with medical experts. Aiming for scalability and robustness, we perform ablation studies comparing mobile-friendly, frame- and video-based architectures and show reliability of the best model by aleatoric and epistemic uncertainty estimates. We hope to pave the road for a community effort toward an accessible, efficient and interpretable screening method and we have started to work on a clinical validation of the proposed method. Data and code are publicly available.

35.Multi-channel MRI Embedding: An EffectiveStrategy for Enhancement of Human Brain WholeTumor Segmentation ⬇️

One of the most important tasks in medical image processing is the brain's whole tumor segmentation. It assists in quicker clinical assessment and early detection of brain tumors, which is crucial for lifesaving treatment procedures of patients. Because, brain tumors often can be malignant or benign, if they are detected at an early stage. A brain tumor is a collection or a mass of abnormal cells in the brain. The human skull encloses the brain very rigidly and any growth inside this restricted place can cause severe health issues. The detection of brain tumors requires careful and intricate analysis for surgical planning and treatment. Most physicians employ Magnetic Resonance Imaging (MRI) to diagnose such tumors. A manual diagnosis of the tumors using MRI is known to be time-consuming; approximately, it takes up to eighteen hours per sample. Thus, the automatic segmentation of tumors has become an optimal solution for this problem. Studies have shown that this technique provides better accuracy and it is faster than manual analysis resulting in patients receiving the treatment at the right time. Our research introduces an efficient strategy called Multi-channel MRI embedding to improve the result of deep learning-based tumor segmentation. The experimental analysis on the Brats-2019 dataset wrt the U-Net encoder-decoder (EnDec) model shows significant improvement. The embedding strategy surmounts the state-of-the-art approaches with an improvement of 2% without any timing overheads.

36.Cosine meets Softmax: A tough-to-beat baseline for visual grounding ⬇️

In this paper, we present a simple baseline for visual grounding for autonomous driving which outperforms the state of the art methods, while retaining minimal design choices. Our framework minimizes the cross-entropy loss over the cosine distance between multiple image ROI features with a text embedding (representing the give sentence/phrase). We use pre-trained networks for obtaining the initial embeddings and learn a transformation layer on top of the text embedding. We perform experiments on the Talk2Car dataset and achieve 68.7% AP50 accuracy, improving upon the previous state of the art by 8.6%. Our investigation suggests reconsideration towards more approaches employing sophisticated attention mechanisms or multi-stage reasoning or complex metric learning loss functions by showing promise in simpler alternatives.

37.Pairwise-GAN: Pose-based View Synthesis through Pair-Wise Training ⬇️

Three-dimensional face reconstruction is one of the popular applications in computer vision. However, even state-of-the-art models still require frontal face as inputs, which restricts its usage scenarios in the wild. A similar dilemma also happens in face recognition. New research designed to recover the frontal face from a single side-pose facial image has emerged. The state-of-the-art in this area is the Face-Transformation generative adversarial network, which is based on the CycleGAN. This inspired our research which explores the performance of two models from pixel transformation in frontal facial synthesis, Pix2Pix and CycleGAN. We conducted the experiments on five different loss functions on Pix2Pix to improve its performance, then followed by proposing a new network Pairwise-GAN in frontal facial synthesis. Pairwise-GAN uses two parallel U-Nets as the generator and PatchGAN as the discriminator. The detailed hyper-parameters are also discussed. Based on the quantitative measurement by face similarity comparison, our results showed that Pix2Pix with L1 loss, gradient difference loss, and identity loss results in 2.72% of improvement at average similarity compared to the default Pix2Pix model. Additionally, the performance of Pairwise-GAN is 5.4% better than the CycleGAN and 9.1% than the Pix2Pix at average similarity.

38.A Review of Visual Descriptors and Classification Techniques Used in Leaf Species Identification ⬇️

Plants are fundamentally important to life. Key research areas in plant science include plant species identification, weed classification using hyper spectral images, monitoring plant health and tracing leaf growth, and the semantic interpretation of leaf information. Botanists easily identify plant species by discriminating between the shape of the leaf, tip, base, leaf margin and leaf vein, as well as the texture of the leaf and the arrangement of leaflets of compound leaves. Because of the increasing demand for experts and calls for biodiversity, there is a need for intelligent systems that recognize and characterize leaves so as to scrutinize a particular species, the diseases that affect them, the pattern of leaf growth, and so on. We review several image processing methods in the feature extraction of leaves, given that feature extraction is a crucial technique in computer vision. As computers cannot comprehend images, they are required to be converted into features by individually analysing image shapes, colours, textures and moments. Images that look the same may deviate in terms of geometric and photometric variations. In our study, we also discuss certain machine learning classifiers for an analysis of different species of leaves.

39.Semantic Segmentation of Surface from Lidar Point Cloud ⬇️

In the field of SLAM (Simultaneous Localization And Mapping) for robot navigation, mapping the environment is an important task. In this regard the Lidar sensor can produce near accurate 3D map of the environment in the format of point cloud, in real time. Though the data is adequate for extracting information related to SLAM, processing millions of points in the point cloud is computationally quite expensive. The methodology presented proposes a fast algorithm that can be used to extract semantically labelled surface segments from the cloud, in real time, for direct navigational use or higher level contextual scene reconstruction. First, a single scan from a spinning Lidar is used to generate a mesh of subsampled cloud points online. The generated mesh is further used for surface normal computation of those points on the basis of which surface segments are estimated. A novel descriptor to represent the surface segments is proposed and utilized to determine the surface class of the segments (semantic label) with the help of classifier. These semantic surface segments can be further utilized for geometric reconstruction of objects in the scene, or can be used for optimized trajectory planning by a robot. The proposed methodology is compared with number of point cloud segmentation methods and state of the art semantic segmentation methods to emphasize its efficacy in terms of speed and accuracy.

40.Calibration Venus: An Interactive Camera Calibration Method Based on Search Algorithm and Pose Decomposition ⬇️

In many scenarios where cameras are applied, such as robot positioning and unmanned driving, camera calibration is one of the most important pre-work. The interactive calibration method based on the plane board is becoming popular in camera calibration field due to its repeatability and operation advantages. However, the existing methods select suggestions from a fixed dataset of pre-defined poses based on subjective experience, which leads to a certain degree of one-sidedness. Moreover, they does not give users clear instructions on how to place the board in the specified pose.

41.Improving Deep Video Compression by Resolution-adaptive Flow Coding ⬇️

In the learning based video compression approaches, it is an essential issue to compress pixel-level optical flow maps by developing new motion vector (MV) encoders. In this work, we propose a new framework called Resolution-adaptive Flow Coding (RaFC) to effectively compress the flow maps globally and locally, in which we use multi-resolution representations instead of single-resolution representations for both the input flow maps and the output motion features of the MV encoder. To handle complex or simple motion patterns globally, our frame-level scheme RaFC-frame automatically decides the optimal flow map resolution for each video frame. To cope different types of motion patterns locally, our block-level scheme called RaFC-block can also select the optimal resolution for each local block of motion features. In addition, the rate-distortion criterion is applied to both RaFC-frame and RaFC-block and select the optimal motion coding mode for effective flow coding. Comprehensive experiments on four benchmark datasets HEVC, VTL, UVG and MCL-JCV clearly demonstrate the effectiveness of our overall RaFC framework after combing RaFC-frame and RaFC-block for video compression.

42.SSKD: Self-Supervised Knowledge Distillation for Cross Domain Adaptive Person Re-Identification ⬇️

Domain adaptive person re-identification (re-ID) is a challenging task due to the large discrepancy between the source domain and the target domain. To reduce the domain discrepancy, existing methods mainly attempt to generate pseudo labels for unlabeled target images by clustering algorithms. However, clustering methods tend to bring noisy labels and the rich fine-grained details in unlabeled images are not sufficiently exploited. In this paper, we seek to improve the quality of labels by capturing feature representation from multiple augmented views of unlabeled images. To this end, we propose a Self-Supervised Knowledge Distillation (SSKD) technique containing two modules, the identity learning and the soft label learning. Identity learning explores the relationship between unlabeled samples and predicts their one-hot labels by clustering to give exact information for confidently distinguished images. Soft label learning regards labels as a distribution and induces an image to be associated with several related classes for training peer network in a self-supervised manner, where the slowly evolving network is a core to obtain soft labels as a gentle constraint for reliable images. Finally, the two modules can resist label noise for re-ID by enhancing each other and systematically integrating label information from unlabeled images. Extensive experiments on several adaptation tasks demonstrate that the proposed method outperforms the current state-of-the-art approaches by large margins.

43.Semi-supervised dictionary learning with graph regularization and active points ⬇️

Supervised Dictionary Learning has gained much interest in the recent decade and has shown significant performance improvements in image classification. However, in general, supervised learning needs a large number of labelled samples per class to achieve an acceptable result. In order to deal with databases which have just a few labelled samples per class, semi-supervised learning, which also exploits unlabelled samples in training phase is used. Indeed, unlabelled samples can help to regularize the learning model, yielding an improvement of classification accuracy. In this paper, we propose a new semi-supervised dictionary learning method based on two pillars: on one hand, we enforce manifold structure preservation from the original data into sparse code space using Locally Linear Embedding, which can be considered a regularization of sparse code; on the other hand, we train a semi-supervised classifier in sparse code space. We show that our approach provides an improvement over state-of-the-art semi-supervised dictionary learning methods.

44.Interpretation of smartphone-captured radiographs utilizing a deep learning-based approach ⬇️

Recently, computer-aided diagnostic systems (CADs) that could automatically interpret medical images effectively have been the emerging subject of recent academic attention. For radiographs, several deep learning-based systems or models have been developed to study the multi-label diseases recognition tasks. However, none of them have been trained to work on smartphone-captured chest radiographs. In this study, we proposed a system that comprises a sequence of deep learning-based neural networks trained on the newly released CheXphoto dataset to tackle this issue. The proposed approach achieved promising results of 0.684 in AUC and 0.699 in average F1 score. To the best of our knowledge, this is the first published study that showed to be capable of processing smartphone-captured radiographs.

45.Synthesizing brain tumor images and annotations by combining progressive growing GAN and SPADE ⬇️

Training segmentation networks requires large annotated datasets, but manual annotation is time consuming and costly. We here investigate if the combination of a noise-to-image GAN and an image-to-image GAN can be used to synthesize realistic brain tumor images as well as the corresponding tumor annotations (labels), to substantially increase the number of training images. The noise-to-image GAN is used to synthesize new label images, while the image-to-image GAN generates the corresponding MR image from the label image. Our results indicate that the two GANs can synthesize label images and MR images that look realistic, and that adding synthetic images improves the segmentation performance, although the effect is small.

46.PolSAR Image Classification Based on Robust Low-Rank Feature Extraction and Markov Random Field ⬇️

Polarimetric synthetic aperture radar (PolSAR) image classification has been investigated vigorously in various remote sensing applications. However, it is still a challenging task nowadays. One significant barrier lies in the speckle effect embedded in the PolSAR imaging process, which greatly degrades the quality of the images and further complicates the classification. To this end, we present a novel PolSAR image classification method, which removes speckle noise via low-rank (LR) feature extraction and enforces smoothness priors via Markov random field (MRF). Specifically, we employ the mixture of Gaussian-based robust LR matrix factorization to simultaneously extract discriminative features and remove complex noises. Then, a classification map is obtained by applying convolutional neural network with data augmentation on the extracted features, where local consistency is implicitly involved, and the insufficient label issue is alleviated. Finally, we refine the classification map by MRF to enforce contextual smoothness. We conduct experiments on two benchmark PolSAR datasets. Experimental results indicate that the proposed method achieves promising classification performance and preferable spatial consistency.

47.Coding Facial Expressions with Gabor Wavelets (IVC Special Issue) ⬇️

We present a method for extracting information about facial expressions from digital images. The method codes facial expression images using a multi-orientation, multi-resolution set of Gabor filters that are topographically ordered and approximately aligned with the face. A similarity space derived from this code is compared with one derived from semantic ratings of the images by human observers. Interestingly the low-dimensional structure of the image-derived similarity space shares organizational features with the circumplex model of affect, suggesting a bridge between categorical and dimensional representations of facial expression. Our results also indicate that it would be possible to construct a facial expression classifier based on a topographically-linked multi-orientation, multi-resolution Gabor coding of the facial images at the input stage. The significant degree of psychological plausibility exhibited by the proposed code may also be useful in the design of human-computer interfaces.

48.Deep Detection for Face Manipulation ⬇️

It has become increasingly challenging to distinguish real faces from their visually realistic fake counterparts, due to the great advances of deep learning based face manipulation techniques in recent years. In this paper, we introduce a deep learning method to detect face manipulation. It consists of two stages: feature extraction and binary classification. To better distinguish fake faces from real faces, we resort to the triplet loss function in the first stage. We then design a simple linear classification network to bridge the learned contrastive features with the real/fake faces. Experimental results on public benchmark datasets demonstrate the effectiveness of this method, and show that it generates better performance than state-of-the-art techniques in most cases.

49.An approach to human iris recognition using quantitative analysis of image features and machine learning ⬇️

The Iris pattern is a unique biological feature for each individual, making it a valuable and powerful tool for human identification. In this paper, an efficient framework for iris recognition is proposed in four steps. (1) Iris segmentation (using a relative total variation combined with Coarse Iris Localization), (2) feature extraction (using Shape&density, FFT, GLCM, GLDM, and Wavelet), (3) feature reduction (employing Kernel-PCA) and (4) classification (applying multi-layer neural network) to classify 2000 iris images of CASIA-Iris-Interval dataset obtained from 200 volunteers. The results confirm that the proposed scheme can provide a reliable prediction with an accuracy of up to 99.64%.

50.A Unified Approach to Kinship Verification ⬇️

In this work, we propose a deep learning-based approach for kin verification using a unified multi-task learning scheme where all kinship classes are jointly learned. This allows us to better utilize small training sets that are typical of kin verification. We introduce a novel approach for fusing the embeddings of kin images, to avoid overfitting, which is a common issue in training such networks. An adaptive sampling scheme is derived for the training set images to resolve the inherent imbalance in kin verification datasets. A thorough ablation study exemplifies the effectivity of our approach, which is experimentally shown to outperform contemporary state-of-the-art kin verification results when applied to the Families In the Wild, FG2018, and FG2020 datasets.

51.Exploring the Hierarchy in Relation Labels for Scene Graph Generation ⬇️

By assigning each relationship a single label, current approaches formulate the relationship detection as a classification problem. Under this formulation, predicate categories are treated as completely different classes. However, different from the object labels where different classes have explicit boundaries, predicates usually have overlaps in their semantic meanings. For example, sit_on and stand_on have common meanings in vertical relationships but different details of how these two objects are vertically placed. In order to leverage the inherent structures of the predicate categories, we propose to first build the language hierarchy and then utilize the Hierarchy Guided Feature Learning (HGFL) strategy to learn better region features of both the coarse-grained level and the fine-grained level. Besides, we also propose the Hierarchy Guided Module (HGM) to utilize the coarse-grained level to guide the learning of fine-grained level features. Experiments show that the proposed simple yet effective method can improve several state-of-the-art baselines by a large margin (up to $33%$ relative gain) in terms of Recall@50 on the task of Scene Graph Generation in different datasets.

52.Map-merging Algorithms for Visual SLAM: Feasibility Study and Empirical Evaluation ⬇️

Simultaneous localization and mapping, especially the one relying solely on video data (vSLAM), is a challenging problem that has been extensively studied in robotics and computer vision. State-of-the-art vSLAM algorithms are capable of constructing accurate-enough maps that enable a mobile robot to autonomously navigate an unknown environment. In this work, we are interested in an important problem related to vSLAM, i.e. map merging, that might appear in various practically important scenarios, e.g. in a multi-robot coverage scenario. This problem asks whether different vSLAM maps can be merged into a consistent single representation. We examine the existing 2D and 3D map-merging algorithms and conduct an extensive empirical evaluation in realistic simulated environment (Habitat). Both qualitative and quantitative comparison is carried out and the obtained results are reported and analyzed.

53.Learning semantic Image attributes using Image recognition and knowledge graph embeddings ⬇️

Extracting structured knowledge from texts has traditionally been used for knowledge base generation. However, other sources of information, such as images can be leveraged into this process to build more complete and richer knowledge bases. Structured semantic representation of the content of an image and knowledge graph embeddings can provide a unique representation of semantic relationships between image entities. Linking known entities in knowledge graphs and learning open-world images using language models has attracted lots of interest over the years. In this paper, we propose a shared learning approach to learn semantic attributes of images by combining a knowledge graph embedding model with the recognized attributes of images. The proposed model premises to help us understand the semantic relationship between the entities of an image and implicitly provide a link for the extracted entities through a knowledge graph embedding model. Under the limitation of using a custom user-defined knowledge base with limited data, the proposed model presents significant accuracy and provides a new alternative to the earlier approaches. The proposed approach is a step towards bridging the gap between frameworks which learn from large amounts of data and frameworks which use a limited set of predicates to infer new knowledge.

54.Revisiting the Threat Space for Vision-based Keystroke Inference Attacks ⬇️

A vision-based keystroke inference attack is a side-channel attack in which an attacker uses an optical device to record users on their mobile devices and infer their keystrokes. The threat space for these attacks has been studied in the past, but we argue that the defining characteristics for this threat space, namely the strength of the attacker, are outdated. Previous works do not study adversaries with vision systems that have been trained with deep neural networks because these models require large amounts of training data and curating such a dataset is expensive. To address this, we create a large-scale synthetic dataset to simulate the attack scenario for a keystroke inference attack. We show that first pre-training on synthetic data, followed by adopting transfer learning techniques on real-life data, increases the performance of our deep learning models. This indicates that these models are able to learn rich, meaningful representations from our synthetic data and that training on the synthetic data can help overcome the issue of having small, real-life datasets for vision-based key stroke inference attacks. For this work, we focus on single keypress classification where the input is a frame of a keypress and the output is a predicted key. We are able to get an accuracy of 95.6% after pre-training a CNN on our synthetic data and training on a small set of real-life data in an adversarial domain adaptation framework. Source Code for Simulator: this https URL

55.A CNN Based Approach for the Near-Field Photometric Stereo Problem ⬇️

Reconstructing the 3D shape of an object using several images under different light sources is a very challenging task, especially when realistic assumptions such as light propagation and attenuation, perspective viewing geometry and specular light reflection are considered. Many of works tackling Photometric Stereo (PS) problems often relax most of the aforementioned assumptions. Especially they ignore specular reflection and global illumination effects. In this work, we propose the first CNN based approach capable of handling these realistic assumptions in Photometric Stereo. We leverage recent improvements of deep neural networks for far-field Photometric Stereo and adapt them to near field setup. We achieve this by employing an iterative procedure for shape estimation which has two main steps. Firstly we train a per-pixel CNN to predict surface normals from reflectance samples. Secondly, we compute the depth by integrating the normal field in order to iteratively estimate light directions and attenuation which is used to compensate the input images to compute reflectance samples for the next iteration. To the best of our knowledge this is the first near-field framework which is able to accurately predict 3D shape from highly specular objects. Our method outperforms competing state-of-the-art near-field Photometric Stereo approaches on both synthetic and real experiments.

56.Micro-Facial Expression Recognition Based on Deep-Rooted Learning Algorithm ⬇️

Facial expressions are important cues to observe human emotions. Facial expression recognition has attracted many researchers for years, but it is still a challenging topic since expression features vary greatly with the head poses, environments, and variations in the different persons involved. In this work, three major steps are involved to improve the performance of micro-facial expression recognition. First, an Adaptive Homomorphic Filtering is used for face detection and rotation rectification processes. Secondly, Micro-facial features were used to extract the appearance variations of a testing image-spatial analysis. The features of motion information are used for expression recognition in a sequence of facial images. An effective Micro-Facial Expression Based Deep-Rooted Learning (MFEDRL) classifier is proposed in this paper to better recognize spontaneous micro-expressions by learning parameters on the optimal features. This proposed method includes two loss functions such as cross entropy loss function and centre loss function. Then the performance of the algorithm will be evaluated using recognition rate and false measures. Simulation results show that the predictive performance of the proposed method outperforms that of the existing classifiers such as Convolutional Neural Network (CNN), Deep Neural Network (DNN), Artificial Neural Network (ANN), Support Vector Machine (SVM), and k-Nearest Neighbours (KNN) in terms of accuracy and Mean Absolute Error (MAE).

57.Removing the Background by Adding the Background: Towards Background Robust Self-supervised Video Representation Learning ⬇️

Self-supervised learning has shown great potentials in improving the video representation ability of deep neural networks by constructing surrogate supervision signals from the unlabeled data. However, some of the current methods tend to suffer from a background cheating problem, i.e., the prediction is highly dependent on the video background instead of the motion, making the model vulnerable to background changes. To alleviate the problem, we propose to remove the background impact by adding the background. That is, given a video, we randomly select a static frame and add it to every other frames to construct a distracting video sample. Then we force the model to pull the feature of the distracting video and the feature of the original video closer, so that the model is explicitly restricted to resist the background influence, focusing more on the motion changes. In addition, in order to prevent the static frame from disturbing the motion area too much, we restrict the feature being consistent with the temporally flipped feature of the reversed video, forcing the model to concentrate more on the motion. We term our method as Temporal-sensitive Background Erasing (TBE). Experiments on UCF101 and HMDB51 show that TBE brings about 6.4% and 4.8% improvements over the state-of-the-art method on the HMDB51 and UCF101 datasets respectively. And it is worth noting that the implementation of our method is so simple and neat and can be added as an additional regularization term to most of the SOTA methods without much efforts.

58.Enhancing Unsupervised Video Representation Learning by Decoupling the Scene and the Motion ⬇️

One significant factor we expect the video representation learning to capture, especially in contrast with the image representation learning, is the object motion. However, we found that in the current mainstream video datasets, some action categories are highly related with the scene where the action happens, making the model tend to degrade to a solution where only the scene information is encoded. For example, a trained model may predict a video as playing football simply because it sees the field, neglecting that the subject is dancing as a cheerleader on the field. This is against our original intention towards the video representation learning and may bring scene bias on different dataset that can not be ignored. In order to tackle this problem, we propose to decouple the scene and the motion (DSM) with two simple operations, so that the model attention towards the motion information is better paid. Specifically, we construct a positive clip and a negative clip for each video. Compared to the original video, the positive/negative is motion-untouched/broken but scene-broken/untouched by Spatial Local Disturbance and Temporal Local Disturbance. Our objective is to pull the positive closer while pushing the negative farther to the original clip in the latent space. In this way, the impact of the scene is weakened while the temporal sensitivity of the network is further enhanced. We conduct experiments on two tasks with various backbones and different pre-training datasets, and find that our method surpass the SOTA methods with a remarkable 8.1% and 8.8% improvement towards action recognition task on the UCF101 and HMDB51 datasets respectively using the same backbone.

59.Multi-Spectral Image Synthesis for Crop/Weed Segmentation in Precision Farming ⬇️

An effective perception system is a fundamental component for farming robots, as it enables them to properly perceive the surrounding environment and to carry out targeted operations. The most recent approaches make use of state-of-the-art machine learning techniques to learn an effective model for the target task. However, those methods need a large amount of labelled data for training. A recent approach to deal with this issue is data augmentation through Generative Adversarial Networks (GANs), where entire synthetic scenes are added to the training data, thus enlarging and diversifying their informative content. In this work, we propose an alternative solution with respect to the common data augmentation techniques, applying it to the fundamental problem of crop/weed segmentation in precision farming. Starting from real images, we create semi-artificial samples by replacing the most relevant object classes (i.e., crop and weeds) with their synthesized counterparts. To do that, we employ a conditional GAN (cGAN), where the generative model is trained by conditioning the shape of the generated object. Moreover, in addition to RGB data, we take into account also near-infrared (NIR) information, generating four channel multi-spectral synthetic images. Quantitative experiments, carried out on three publicly available datasets, show that (i) our model is capable of generating realistic multi-spectral images of plants and (ii) the usage of such synthetic images in the training process improves the segmentation performance of state-of-the-art semantic segmentation Convolutional Networks.

60.Smoothness Sensor: Adaptive Smoothness-Transition Graph Convolutions for Attributed Graph Clustering ⬇️

Clustering techniques attempt to group objects with similar properties into a cluster. Clustering the nodes of an attributed graph, in which each node is associated with a set of feature attributes, has attracted significant attention. Graph convolutional networks (GCNs) represent an effective approach for integrating the two complementary factors of node attributes and structural information for attributed graph clustering. However, oversmoothing of GCNs produces indistinguishable representations of nodes, such that the nodes in a graph tend to be grouped into fewer clusters, and poses a challenge due to the resulting performance drop. In this study, we propose a smoothness sensor for attributed graph clustering based on adaptive smoothness-transition graph convolutions, which senses the smoothness of a graph and adaptively terminates the current convolution once the smoothness is saturated to prevent oversmoothing. Furthermore, as an alternative to graph-level smoothness, a novel fine-gained node-wise level assessment of smoothness is proposed, in which smoothness is computed in accordance with the neighborhood conditions of a given node at a certain order of graph convolution. In addition, a self-supervision criterion is designed considering both the tightness within clusters and the separation between clusters to guide the whole neural network training process. Experiments show that the proposed methods significantly outperform 12 other state-of-the-art baselines in terms of three different metrics across four benchmark datasets. In addition, an extensive study reveals the reasons for their effectiveness and efficiency.

61.Monitoring Spatial Sustainable Development: semi-automated analysis of Satellite and Aerial Images for Energy Transition and Sustainability Indicators ⬇️

This report presents the results of the DeepSolaris project that was carried out under the ESS action 'Merging Geostatistics and Geospatial Information in Member States'. During the project several deep learning algorithms were evaluated to detect solar panels in remote sensing data. The aim of the project was to evaluate whether deep learning models could be developed, that worked across different member states in the European Union. Two remote sensing data sources were considered: aerial images on the one hand, and satellite images on the other. Two flavours of deep learning models were evaluated: classification models and object detection models. For the evaluation of the deep learning models we used a cross-site evaluation approach: the deep learning models where trained in one geographical area and then evaluated on a different geographical area, previously unseen by the algorithm. The cross-site evaluation was furthermore carried out twice: deep learning models trained on he Netherlands were evaluated on Germany and vice versa. While the deep learning models were able to detect solar panels successfully, false detection remained a problem. Moreover, model performance decreased dramatically when evaluated in a cross-border fashion. Hence, training a model that performs reliably across different countries in the European Union is a challenging task. That being said, the models detected quite a share of solar panels not present in current solar panel registers and therefore can already be used as-is to help reduced manual labor in checking these registers.

62.Abstractive Information Extraction from Scanned Invoices (AIESI) using End-to-end Sequential Approach ⬇️

Recent proliferation in the field of Machine Learning and Deep Learning allows us to generate OCR models with higher accuracy. Optical Character Recognition(OCR) is the process of extracting text from documents and scanned images. For document data streamlining, we are interested in data like, Payee name, total amount, address, and etc. Extracted information helps to get complete insight of data, which can be helpful for fast document searching, efficient indexing in databases, data analytics, and etc. Using AIESI we can eliminate human effort for key parameters extraction from scanned documents. Abstract Information Extraction from Scanned Invoices (AIESI) is a process of extracting information like, date, total amount, payee name, and etc from scanned receipts. In this paper we proposed an improved method to ensemble all visual and textual features from invoices to extract key invoice parameters using Word wise BiLSTM.

63.Generator Versus Segmentor: Pseudo-healthy Synthesis ⬇️

Pseudo-healthy synthesis is defined as synthesizing a subject-specific 'healthy' image from a pathological one, with applications ranging from segmentation to anomaly detection. In recent years, the existing GAN-based methods proposed for pseudo-healthy synthesis aim to eliminate the global differences between synthetic and healthy images. In this paper, we discuss the problems of these approaches, which are the style transfer and artifacts respectively. To address these problems, we consider the local differences between the lesions and normal tissue. To achieve this, we propose an adversarial training regime that alternatively trains a generator and a segmentor. The segmentor is trained to distinguish the synthetic lesions (i.e. the region in synthetic images corresponding to the lesions in the pathological ones) from the normal tissue, while the generator is trained to deceive the segmentor by transforming lesion regions into lesion-free-like ones and preserve the normal tissue at the same time. Qualitative and quantitative experimental results on public datasets BraTS and LiTS demonstrate that the proposed method outperforms state-of-the-art methods by preserving style and removing the artifacts. Our implementation is publicly available at this https URL

64.Short-Term and Long-Term Context Aggregation Network for Video Inpainting ⬇️

Video inpainting aims to restore missing regions of a video and has many applications such as video editing and object removal. However, existing methods either suffer from inaccurate short-term context aggregation or rarely explore long-term frame information. In this work, we present a novel context aggregation network to effectively exploit both short-term and long-term frame information for video inpainting. In the encoding stage, we propose boundary-aware short-term context aggregation, which aligns and aggregates, from neighbor frames, local regions that are closely related to the boundary context of missing regions into the target frame. Furthermore, we propose dynamic long-term context aggregation to globally refine the feature map generated in the encoding stage using long-term frame features, which are dynamically updated throughout the inpainting process. Experiments show that it outperforms state-of-the-art methods with better inpainting results and fast inpainting speed.

65.YOLObile: Real-Time Object Detection on Mobile Devices via Compression-Compilation Co-Design ⬇️

The rapid development and wide utilization of object detection techniques have aroused attention on both accuracy and speed of object detectors. However, the current state-of-the-art object detection works are either accuracy-oriented using a large model but leading to high latency or speed-oriented using a lightweight model but sacrificing accuracy. In this work, we propose YOLObile framework, a real-time object detection on mobile devices via compression-compilation co-design. A novel block-punched pruning scheme is proposed for any kernel size. To improve computational efficiency on mobile devices, a GPU-CPU collaborative scheme is adopted along with advanced compiler-assisted optimizations. Experimental results indicate that our pruning scheme achieves 14$\times$ compression rate of YOLOv4 with 49.0 mAP. Under our YOLObile framework, we achieve 17 FPS inference speed using GPU on Samsung Galaxy S20. By incorporating our proposed GPU-CPU collaborative scheme, the inference speed is increased to 19.1 FPS, and outperforms the original YOLOv4 by 5$\times$ speedup.

66.RGB2LIDAR: Towards Solving Large-Scale Cross-Modal Visual Localization ⬇️

We study an important, yet largely unexplored problem of large-scale cross-modal visual localization by matching ground RGB images to a geo-referenced aerial LIDAR 3D point cloud (rendered as depth images). Prior works were demonstrated on small datasets and did not lend themselves to scaling up for large-scale applications. To enable large-scale evaluation, we introduce a new dataset containing over 550K pairs (covering 143 km^2 area) of RGB and aerial LIDAR depth images. We propose a novel joint embedding based method that effectively combines the appearance and semantic cues from both modalities to handle drastic cross-modal variations. Experiments on the proposed dataset show that our model achieves a strong result of a median rank of 5 in matching across a large test set of 50K location pairs collected from a 14km^2 area. This represents a significant advancement over prior works in performance and scale. We conclude with qualitative results to highlight the challenging nature of this task and the benefits of the proposed model. Our work provides a foundation for further research in cross-modal visual localization.

67.AttnGrounder: Talking to Cars with Attention ⬇️

We propose Attention Grounder (AttnGrounder), a single-stage end-to-end trainable model for the task of visual grounding. Visual grounding aims to localize a specific object in an image based on a given natural language text query. Unlike previous methods that use the same text representation for every image region, we use a visual-text attention module that relates each word in the given query with every region in the corresponding image for constructing a region dependent text representation. Furthermore, for improving the localization ability of our model, we use our visual-text attention module to generate an attention mask around the referred object. The attention mask is trained as an auxiliary task using a rectangular mask generated with the provided ground-truth coordinates. We evaluate AttnGrounder on the Talk2Car dataset and show an improvement of 3.26% over the existing methods.

68.A Progressive Sub-Network Searching Framework for Dynamic Inference ⬇️

Many techniques have been developed, such as model compression, to make Deep Neural Networks (DNNs) inference more efficiently. Nevertheless, DNNs still lack excellent run-time dynamic inference capability to enable users trade-off accuracy and computation complexity (i.e., latency on target hardware) after model deployment, based on dynamic requirements and environments. Such research direction recently draws great attention, where one realization is to train the target DNN through a multiple-term objective function, which consists of cross-entropy terms from multiple sub-nets. Our investigation in this work show that the performance of dynamic inference highly relies on the quality of sub-net sampling. With objective to construct a dynamic DNN and search multiple high quality sub-nets with minimal searching cost, we propose a progressive sub-net searching framework, which is embedded with several effective techniques, including trainable noise ranking, channel group and fine-tuning threshold setting, sub-nets re-selection. The proposed framework empowers the target DNN with better dynamic inference capability, which outperforms prior works on both CIFAR-10 and ImageNet dataset via comprehensive experiments on different network structures. Taken ResNet18 as an example, our proposed method achieves much better dynamic inference accuracy compared with prior popular Universally-Slimmable-Network by 4.4%-maximally and 2.3%-averagely in ImageNet dataset with the same model size.

69.Inverse mapping of face GANs ⬇️

Generative adversarial networks (GANs) synthesize realistic images from a random latent vector. While many studies have explored various training configurations and architectures for GANs, the problem of inverting a generative model to extract latent vectors of given input images has been inadequately investigated. Although there is exactly one generated image per given random vector, the mapping from an image to its recovered latent vector can have more than one solution. We train a ResNet architecture to recover a latent vector for a given face that can be used to generate a face nearly identical to the target. We use a perceptual loss to embed face details in the recovered latent vector while maintaining visual quality using a pixel loss. The vast majority of studies on latent vector recovery perform well only on generated images, we argue that our method can be used to determine a mapping between real human faces and latent-space vectors that contain most of the important face style details. In addition, our proposed method projects generated faces to their latent-space with high fidelity and speed. At last, we demonstrate the performance of our approach on both real and generated faces.

70.KSM: Fast Multiple Task Adaption via Kernel-wise Soft Mask Learning ⬇️

Deep Neural Networks (DNN) could forget the knowledge about earlier tasks when learning new tasks, and this is known as \textit{catastrophic forgetting}. While recent continual learning methods are capable of alleviating the catastrophic problem on toy-sized datasets, some issues still remain to be tackled when applying them in real-world problems. Recently, the fast mask-based learning method (e.g. piggyback \cite{mallya2018piggyback}) is proposed to address these issues by learning only a binary element-wise mask in a fast manner, while keeping the backbone model fixed. However, the binary mask has limited modeling capacity for new tasks. A more recent work \cite{hung2019compacting} proposes a compress-grow-based method (CPG) to achieve better accuracy for new tasks by partially training backbone model, but with order-higher training cost, which makes it infeasible to be deployed into popular state-of-the-art edge-/mobile-learning. The primary goal of this work is to simultaneously achieve fast and high-accuracy multi task adaption in continual learning setting. Thus motivated, we propose a new training method called \textit{kernel-wise Soft Mask} (KSM), which learns a kernel-wise hybrid binary and real-value soft mask for each task, while using the same backbone model. Such a soft mask can be viewed as a superposition of a binary mask and a properly scaled real-value tensor, which offers a richer representation capability without low-level kernel support to meet the objective of low hardware overhead. We validate KSM on multiple benchmark datasets against recent state-of-the-art methods (e.g. Piggyback, Packnet, CPG, etc.), which shows good improvement in both accuracy and training cost.

71.Deep Hiearchical Multi-Label Classification Applied to Chest X-Ray Abnormality Taxonomies ⬇️

CXRs are a crucial and extraordinarily common diagnostic tool, leading to heavy research for CAD solutions. However, both high classification accuracy and meaningful model predictions that respect and incorporate clinical taxonomies are crucial for CAD usability. To this end, we present a deep HMLC approach for CXR CAD. Different than other hierarchical systems, we show that first training the network to model conditional probability directly and then refining it with unconditional probabilities is key in boosting performance. In addition, we also formulate a numerically stable cross-entropy loss function for unconditional probabilities that provides concrete performance improvements. Finally, we demonstrate that HMLC can be an effective means to manage missing or incomplete labels. To the best of our knowledge, we are the first to apply HMLC to medical imaging CAD. We extensively evaluate our approach on detecting abnormality labels from the CXR arm of the PLCO dataset, which comprises over $198,000$ manually annotated CXRs. When using complete labels, we report a mean AUC of 0.887, the highest yet reported for this dataset. These results are supported by ancillary experiments on the PadChest dataset, where we also report significant improvements, 1.2% and 4.1% in AUC and AP, respectively over strong "flat" classifiers. Finally, we demonstrate that our HMLC approach can much better handle incompletely labelled data. These performance improvements, combined with the inherent usefulness of taxonomic predictions, indicate that our approach represents a useful step forward for CXR CAD.

72.3D Reconstruction and Segmentation of Dissection Photographs for MRI-free Neuropathology ⬇️

Neuroimaging to neuropathology correlation (NTNC) promises to enable the transfer of microscopic signatures of pathology to in vivo imaging with MRI, ultimately enhancing clinical care. NTNC traditionally requires a volumetric MRI scan, acquired either ex vivo or a short time prior to death. Unfortunately, ex vivo MRI is difficult and costly, and recent premortem scans of sufficient quality are seldom available. To bridge this gap, we present methodology to 3D reconstruct and segment full brain image volumes from brain dissection photographs, which are routinely acquired at many brain banks and neuropathology departments. The 3D reconstruction is achieved via a joint registration framework, which uses a reference volume other than MRI. This volume may represent either the sample at hand (e.g., a surface 3D scan) or the general population (a probabilistic atlas). In addition, we present a Bayesian method to segment the 3D reconstructed photographic volumes into 36 neuroanatomical structures, which is robust to nonuniform brightness within and across photographs. We evaluate our methods on a dataset with 24 brains, using Dice scores and volume correlations. The results show that dissection photography is a valid replacement for ex vivo MRI in many volumetric analyses, opening an avenue for MRI-free NTNC, including retrospective data. The code is available at this https URL.

73.Label-Free Segmentation of COVID-19 Lesions in Lung CT ⬇️

Scarcity of annotated images hampers the building of automated solution for reliable COVID-19 diagnosis and evaluation from CT. To alleviate the burden of data annotation, we herein present a label-free approach for segmenting COVID-19 lesions in CT via pixel-level anomaly modeling that mines out the relevant knowledge from normal CT lung scans. Our modeling is inspired by the observation that the parts of tracheae and vessels, which lay in the high-intensity range where lesions belong to, exhibit strong patterns. To facilitate the learning of such patterns at a pixel level, we synthesize lesions' using a set of surprisingly simple operations and insert the synthesized lesions' into normal CT lung scans to form training pairs, from which we learn a normalcy-converting network (NormNet) that turns an 'abnormal' image back to normal. Our experiments on three different datasets validate the effectiveness of NormNet, which conspicuously outperforms a variety of unsupervised anomaly detection (UAD) methods.

74.Automatic elimination of the pectoral muscle in mammograms based on anatomical features ⬇️

Digital mammogram inspection is the most popular technique for early detection of abnormalities in human breast tissue. When mammograms are analyzed through a computational method, the presence of the pectoral muscle might affect the results of breast lesions detection. This problem is particularly evident in the mediolateral oblique view (MLO), where pectoral muscle occupies a large part of the mammography. Therefore, identifying and eliminating the pectoral muscle are essential steps for improving the automatic discrimination of breast tissue. In this paper, we propose an approach based on anatomical features to tackle this problem. Our method consists of two steps: (1) a process to remove the noisy elements such as labels, markers, scratches and wedges, and (2) application of an intensity transformation based on the Beta distribution. The novel methodology is tested with 322 digital mammograms from the Mammographic Image Analysis Society (mini-MIAS) database and with a set of 84 mammograms for which the area normalized error was previously calculated. The results show a very good performance of the method.

75.A Multisensory Learning Architecture for Rotation-invariant Object Recognition ⬇️

This study presents a multisensory machine learning architecture for object recognition by employing a novel dataset that was constructed with the iCub robot, which is equipped with three cameras and a depth sensor. The proposed architecture combines convolutional neural networks to form representations (i.e., features) for grayscaled color images and a multi-layer perceptron algorithm to process depth data. To this end, we aimed to learn joint representations of different modalities (e.g., color and depth) and employ them for recognizing objects. We evaluate the performance of the proposed architecture by benchmarking the results obtained with the models trained separately with the input of different sensors and a state-of-the-art data fusion technique, namely decision level fusion. The results show that our architecture improves the recognition accuracy compared with the models that use inputs from a single modality and decision level multimodal fusion method.

76.VC-Net: Deep Volume-Composition Networks for Segmentation and Visualization of Highly Sparse and Noisy Image Data ⬇️

The motivation of our work is to present a new visualization-guided computing paradigm to combine direct 3D volume processing and volume rendered clues for effective 3D exploration such as extracting and visualizing microstructures in-vivo. However, it is still challenging to extract and visualize high fidelity 3D vessel structure due to its high sparseness, noisiness, and complex topology variations. In this paper, we present an end-to-end deep learning method, VC-Net, for robust extraction of 3D microvasculature through embedding the image composition, generated by maximum intensity projection (MIP), into 3D volume image learning to enhance the performance. The core novelty is to automatically leverage the volume visualization technique (MIP) to enhance the 3D data exploration at deep learning level. The MIP embedding features can enhance the local vessel signal and are adaptive to the geometric variability and scalability of vessels, which is crucial in microvascular tracking. A multi-stream convolutional neural network is proposed to learn the 3D volume and 2D MIP features respectively and then explore their inter-dependencies in a joint volume-composition embedding space by unprojecting the MIP features into 3D volume embedding space. The proposed framework can better capture small / micro vessels and improve vessel connectivity. To our knowledge, this is the first deep learning framework to construct a joint convolutional embedding space, where the computed vessel probabilities from volume rendering based 2D projection and 3D volume can be explored and integrated synergistically. Experimental results are compared with the traditional 3D vessel segmentation methods and the deep learning state-of-the-art on public and real patient (micro-)cerebrovascular image datasets. Our method demonstrates the potential in a powerful MR arteriogram and venogram diagnosis of vascular diseases.

77.Mathematical Morphology via Category Theory ⬇️

Mathematical morphology contributes many profitable tools to image processing area. Some of these methods considered to be basic but the most important fundamental of data processing in many various applications. In this paper, we modify the fundamental of morphological operations such as dilation and erosion making use of limit and co-limit preserving functors within (Category Theory). Adopting the well-known matrix representation of images, the category of matrix, called Mat, can be represented as an image. With enriching Mat over various semirings such as Boolean and (max,+) semirings, one can arrive at classical definition of binary and gray-scale images using the categorical tensor product in Mat. With dilation operation in hand, the erosion can be reached using the famous tensor-hom adjunction. This approach enables us to define new types of dilation and erosion between two images represented by matrices using different semirings other than Boolean and (max,+) semirings. The viewpoint of morphological operations from category theory can also shed light to the claimed concept that mathematical morphology is a model for linear logic.

78.Towards the Quantification of Safety Risks in Deep Neural Networks ⬇️

Safety concerns on the deep neural networks (DNNs) have been raised when they are applied to critical sectors. In this paper, we define safety risks by requesting the alignment of the network's decision with human perception. To enable a general methodology for quantifying safety risks, we define a generic safety property and instantiate it to express various safety risks. For the quantification of risks, we take the maximum radius of safe norm balls, in which no safety risk exists. The computation of the maximum safe radius is reduced to the computation of their respective Lipschitz metrics - the quantities to be computed. In addition to the known adversarial example, reachability example, and invariant example, in this paper we identify a new class of risk - uncertainty example - on which humans can tell easily but the network is unsure. We develop an algorithm, inspired by derivative-free optimization techniques and accelerated by tensor-based parallelization on GPUs, to support efficient computation of the metrics. We perform evaluations on several benchmark neural networks, including ACSC-Xu, MNIST, CIFAR-10, and ImageNet networks. The experiments show that, our method can achieve competitive performance on safety quantification in terms of the tightness and the efficiency of computation. Importantly, as a generic approach, our method can work with a broad class of safety risks and without restrictions on the structure of neural networks.

79.Extracting Optimal Solution Manifolds using Constrained Neural Optimization ⬇️

Constrained Optimization solution algorithms are restricted to point based solutions. In practice, single or multiple objectives must be satisfied, wherein both the objective function and constraints can be non-convex resulting in multiple optimal solutions. Real world scenarios include intersecting surfaces as Implicit Functions, Hyperspectral Unmixing and Pareto Optimal fronts. Local or global convexification is a common workaround when faced with non-convex forms. However, such an approach is often restricted to a strict class of functions, deviation from which results in sub-optimal solution to the original problem. We present neural solutions for extracting optimal sets as approximate manifolds, where unmodified, non-convex objectives and constraints are defined as modeler guided, domain-informed $L_2$ loss function. This promotes interpretability since modelers can confirm the results against known analytical forms in their specific domains. We present synthetic and realistic cases to validate our approach and compare against known solvers for bench-marking in terms of accuracy and computational efficiency.

80.Attention Cube Network for Image Restoration ⬇️

Recently, deep convolutional neural network (CNN) have been widely used in image restoration and obtained great success. However, most of existing methods are limited to local receptive field and equal treatment of different types of information. Besides, existing methods always use a multi-supervised method to aggregate different feature maps, which can not effectively aggregate hierarchical feature information. To address these issues, we propose an attention cube network (A-CubeNet) for image restoration for more powerful feature expression and feature correlation learning. Specifically, we design a novel attention mechanism from three dimensions, namely spatial dimension, channel-wise dimension and hierarchical dimension. The adaptive spatial attention branch (ASAB) and the adaptive channel attention branch (ACAB) constitute the adaptive dual attention module (ADAM), which can capture the long-range spatial and channel-wise contextual information to expand the receptive field and distinguish different types of information for more effective feature representations. Furthermore, the adaptive hierarchical attention module (AHAM) can capture the long-range hierarchical contextual information to flexibly aggregate different feature maps by weights depending on the global context. The ADAM and AHAM cooperate to form an "attention in attention" structure, which means AHAM's inputs are enhanced by ASAB and ACAB. Experiments demonstrate the superiority of our method over state-of-the-art image restoration methods in both quantitative comparison and visual analysis.

81.How Much Can We Really Trust You? Towards Simple, Interpretable Trust Quantification Metrics for Deep Neural Networks ⬇️

A critical step to building trustworthy deep neural networks is trust quantification, where we ask the question: How much can we trust a deep neural network? In this study, we take a step towards simple, interpretable metrics for trust quantification by introducing a suite of metrics for assessing the overall trustworthiness of deep neural networks based on their behaviour when answering a set of questions. We conduct a thought experiment and explore two key questions about trust in relation to confidence: 1) How much trust do we have in actors who give wrong answers with great confidence?, and 2) How much trust do we have in actors who give right answers hesitantly? Based on insights gained, we introduce the concept of question-answer trust to quantify trustworthiness of an individual answer based on confident behaviour under correct and incorrect answer scenarios, and the concept of trust density to characterize the distribution of overall trust for an individual answer scenario. We further introduce the concept of trust spectrum for representing overall trust with respect to the spectrum of possible answer scenarios across correctly and incorrectly answered questions. Finally, we introduce NetTrustScore, a scalar metric summarizing overall trustworthiness. The suite of metrics aligns with past social psychology studies that study the relationship between trust and confidence. Leveraging these metrics, we quantify the trustworthiness of several well-known deep neural network architectures for image recognition to get a deeper understanding of where trust breaks down. The proposed metrics are by no means perfect, but the hope is to push the conversation towards better metrics to help guide practitioners and regulators in producing, deploying, and certifying deep learning solutions that can be trusted to operate in real-world, mission-critical scenarios.

82.Multi-Channel Potts-Based Reconstruction for Multi-Spectral Computed Tomography ⬇️

We consider reconstructing multi-channel images from measurements performed by photon-counting and energy-discriminating detectors in the setting of multi-spectral X-ray computed tomography (CT). Our aim is to exploit the strong structural correlation that is known to exist between the channels of multi-spectral CT images. To that end, we adopt the multi-channel Potts prior to jointly reconstruct all channels. This prior produces piecewise constant solutions with strongly correlated channels. In particular, edges are enforced to have the same spatial position across channels which is a benefit over TV-based methods. We consider the Potts prior in two frameworks: (a) in the context of a variational Potts model, and (b) in a Potts-superiorization approach that perturbs the iterates of a basic iterative least squares solver. We identify an alternating direction method of multipliers (ADMM) approach as well as a Potts-superiorized conjugate gradient method as particularly suitable. In numerical experiments, we compare the Potts prior based approaches to existing TV-type approaches on realistically simulated multi-spectral CT data and obtain improved reconstruction for compound solid bodies.

83.Segmentation of Lungs in Chest X-Ray Image Using Generative Adversarial Networks ⬇️

Chest X-ray (CXR) is a low-cost medical imaging technique. It is a common procedure for the identification of many respiratory diseases compared to MRI, CT, and PET scans. This paper presents the use of generative adversarial networks (GAN) to perform the task of lung segmentation on a given CXR. GANs are popular to generate realistic data by learning the mapping from one domain to another. In our work, the generator of the GAN is trained to generate a segmented mask of a given input CXR. The discriminator distinguishes between a ground truth and the generated mask, and updates the generator through the adversarial loss measure. The objective is to generate masks for the input CXR, which are as realistic as possible compared to the ground truth masks. The model is trained and evaluated using four different discriminators referred to as D1, D2, D3, and D4, respectively. Experimental results on three different CXR datasets reveal that the proposed model is able to achieve a dice-score of 0.9740, and IOU score of 0.943, which are better than other reported state-of-the art results.

84.Efficient Folded Attention for 3D Medical Image Reconstruction and Segmentation ⬇️

Recently, 3D medical image reconstruction (MIR) and segmentation (MIS) based on deep neural networks have been developed with promising results, and attention mechanism has been further designed to capture global contextual information for performance enhancement. However, the large size of 3D volume images poses a great computational challenge to traditional attention methods. In this paper, we propose a folded attention (FA) approach to improve the computational efficiency of traditional attention methods on 3D medical images. The main idea is that we apply tensor folding and unfolding operations with four permutations to build four small sub-affinity matrices to approximate the original affinity matrix. Through four consecutive sub-attention modules of FA, each element in the feature tensor can aggregate spatial-channel information from all other elements. Compared to traditional attention methods, with moderate improvement of accuracy, FA can substantially reduce the computational complexity and GPU memory consumption. We demonstrate the superiority of our method on two challenging tasks for 3D MIR and MIS, which are quantitative susceptibility mapping and multiple sclerosis lesion segmentation.