1.RISP: Rendering-Invariant State Predictor with Differentiable Simulation and Rendering for Cross-Domain Parameter Estimation ⬇️
This work considers identifying parameters characterizing a physical system's dynamic motion directly from a video whose rendering configurations are inaccessible. Existing solutions require massive training data or lack generalizability to unknown rendering configurations. We propose a novel approach that marries domain randomization and differentiable rendering gradients to address this problem. Our core idea is to train a rendering-invariant state-prediction (RISP) network that transforms image differences into state differences independent of rendering configurations, e.g., lighting, shadows, or material reflectance. To train this predictor, we formulate a new loss on rendering variances using gradients from differentiable rendering. Moreover, we present an efficient, second-order method to compute the gradients of this loss, allowing it to be integrated seamlessly into modern deep learning frameworks. We evaluate our method in rigid-body and deformable-body simulation environments using four tasks: state estimation, system identification, imitation learning, and visuomotor control. We further demonstrate the efficacy of our approach on a real-world example: inferring the state and action sequences of a quadrotor from a video of its motion sequences. Compared with existing methods, our approach achieves significantly lower reconstruction errors and has better generalizability among unknown rendering configurations.
2.HULC: 3D Human Motion Capture with Pose Manifold Sampling and Dense Contact Guidance ⬇️
Marker-less monocular 3D human motion capture (MoCap) with scene interactions is a challenging research topic relevant for extended reality, robotics and virtual avatar generation. Due to the inherent depth ambiguity of monocular settings, 3D motions captured with existing methods often contain severe artefacts such as incorrect body-scene inter-penetrations, jitter and body floating. To tackle these issues, we propose HULC, a new approach for 3D human MoCap which is aware of the scene geometry. HULC estimates 3D poses and dense body-environment surface contacts for improved 3D localisations, as well as the absolute scale of the subject. Furthermore, we introduce a 3D pose trajectory optimisation based on a novel pose manifold sampling that resolves erroneous body-environment inter-penetrations. Although the proposed method requires less structured inputs compared to existing scene-aware monocular MoCap algorithms, it produces more physically-plausible poses: HULC significantly and consistently outperforms the existing approaches in various experiments and on different metrics.
3.Revisiting Random Channel Pruning for Neural Network Compression ⬇️
Channel (or 3D filter) pruning serves as an effective way to accelerate the inference of neural networks. There has been a flurry of algorithms that try to solve this practical problem, each being claimed effective in some ways. Yet, a benchmark to compare those algorithms directly is lacking, mainly due to the complexity of the algorithms and some custom settings such as the particular network configuration or training procedure. A fair benchmark is important for the further development of channel pruning.
Meanwhile, recent investigations reveal that the channel configurations discovered by pruning algorithms are at least as important as the pre-trained weights. This gives channel pruning a new role, namely searching the optimal channel configuration. In this paper, we try to determine the channel configuration of the pruned models by random search. The proposed approach provides a new way to compare different methods, namely how well they behave compared with random pruning. We show that this simple strategy works quite well compared with other channel pruning methods. We also show that under this setting, there are surprisingly no clear winners among different channel importance evaluation methods, which then may tilt the research efforts into advanced channel configuration searching methods.
4.NTIRE 2022 Challenge on Efficient Super-Resolution: Methods and Results ⬇️
This paper reviews the NTIRE 2022 challenge on efficient single image super-resolution with focus on the proposed solutions and results. The task of the challenge was to super-resolve an input image with a magnification factor of $\times$4 based on pairs of low and corresponding high resolution images. The aim was to design a network for single image super-resolution that achieved improvement of efficiency measured according to several metrics including runtime, parameters, FLOPs, activations, and memory consumption while at least maintaining the PSNR of 29.00dB on DIV2K validation set. IMDN is set as the baseline for efficiency measurement. The challenge had 3 tracks including the main track (runtime), sub-track one (model complexity), and sub-track two (overall performance). In the main track, the practical runtime performance of the submissions was evaluated. The rank of the teams were determined directly by the absolute value of the average runtime on the validation set and test set. In sub-track one, the number of parameters and FLOPs were considered. And the individual rankings of the two metrics were summed up to determine a final ranking in this track. In sub-track two, all of the five metrics mentioned in the description of the challenge including runtime, parameter count, FLOPs, activations, and memory consumption were considered. Similar to sub-track one, the rankings of five metrics were summed up to determine a final ranking. The challenge had 303 registered participants, and 43 teams made valid submissions. They gauge the state-of-the-art in efficient single image super-resolution.
5.RepSR: Training Efficient VGG-style Super-Resolution Networks with Structural Re-Parameterization and Batch Normalization ⬇️
This paper explores training efficient VGG-style super-resolution (SR) networks with the structural re-parameterization technique. The general pipeline of re-parameterization is to train networks with multi-branch topology first, and then merge them into standard 3x3 convolutions for efficient inference. In this work, we revisit those primary designs and investigate essential components for re-parameterizing SR networks. First of all, we find that batch normalization (BN) is important to bring training non-linearity and improve the final performance. However, BN is typically ignored in SR, as it usually degrades the performance and introduces unpleasant artifacts. We carefully analyze the cause of BN issue and then propose a straightforward yet effective solution. In particular, we first train SR networks with mini-batch statistics as usual, and then switch to using population statistics at the later training period. While we have successfully re-introduced BN into SR, we further design a new re-parameterizable block tailored for SR, namely RepSR. It consists of a clean residual path and two expand-and-squeeze convolution paths with the modified BN. Extensive experiments demonstrate that our simple RepSR is capable of achieving superior performance to previous SR re-parameterization methods among different model sizes. In addition, our RepSR can achieve a better trade-off between performance and actual running time (throughput) than previous SR methods. Codes will be available at this https URL.
6.Video-ReTime: Learning Temporally Varying Speediness for Time Remapping ⬇️
We propose a method for generating a temporally remapped video that matches the desired target duration while maximally preserving natural video dynamics. Our approach trains a neural network through self-supervision to recognize and accurately localize temporally varying changes in the video playback speed. To re-time videos, we 1. use the model to infer the slowness of individual video frames, and 2. optimize the temporal frame sub-sampling to be consistent with the model's slowness predictions. We demonstrate that this model can detect playback speed variations more accurately while also being orders of magnitude more efficient than prior approaches. Furthermore, we propose an optimization for video re-timing that enables precise control over the target duration and performs more robustly on longer videos than prior methods. We evaluate the model quantitatively on artificially speed-up videos, through transfer to action recognition, and qualitatively through user studies.
7.TDT: Teaching Detectors to Track without Fully Annotated Videos ⬇️
Recently, one-stage trackers that use a joint model to predict both detections and appearance embeddings in one forward pass received much attention and achieved state-of-the-art results on the Multi-Object Tracking (MOT) benchmarks. However, their success depends on the availability of videos that are fully annotated with tracking data, which is expensive and hard to obtain. This can limit the model generalization. In comparison, the two-stage approach, which performs detection and embedding separately, is slower but easier to train as their data are easier to annotate. We propose to combine the best of the two worlds through a data distillation approach. Specifically, we use a teacher embedder, trained on Re-ID datasets, to generate pseudo appearance embedding labels for the detection datasets. Then, we use the augmented dataset to train a detector that is also capable of regressing these pseudo-embeddings in a fully-convolutional fashion. Our proposed one-stage solution matches the two-stage counterpart in quality but is 3 times faster. Even though the teacher embedder has not seen any tracking data during training, our proposed tracker achieves competitive performance with some popular trackers (e.g. JDE) trained with fully labeled tracking data.
8.Face Detection on Mobile: Five Implementations and Analysis ⬇️
In many practical cases face detection on smartphones or other highly portable devices is a necessity. Applications include mobile face access control systems, driver status tracking, emotion recognition, etc. Mobile devices have limited processing power and should have long-enough battery life even with face detection application running. Thus, striking the right balance between algorithm quality and complexity is crucial. In this work we adapt 5 algorithms to mobile. These algorithms are based on handcrafted or neural-network-based features and include: Viola-Jones (Haar cascade), LBP, HOG, MTCNN, BlazeFace. We analyze inference time of these algorithms on different devices with different input image resolutions. We provide guidance, which algorithms are the best fit for mobile face access control systems and potentially other mobile applications. Interestingly, we note that cascaded algorithms perform faster on scenes without faces, while BlazeFace is slower on empty scenes. Exploiting this behavior might be useful in practice.
9.Review on Panoramic Imaging and Its Applications in Scene Understanding ⬇️
With the rapid development of high-speed communication and artificial intelligence technologies, human perception of real-world scenes is no longer limited to the use of small Field of View (FoV) and low-dimensional scene detection devices. Panoramic imaging emerges as the next generation of innovative intelligent instruments for environmental perception and measurement. However, while satisfying the need for large-FoV photographic imaging, panoramic imaging instruments are expected to have high resolution, no blind area, miniaturization, and multi-dimensional intelligent perception, and can be combined with artificial intelligence methods towards the next generation of intelligent instruments, enabling deeper understanding and more holistic perception of 360-degree real-world surrounding environments. Fortunately, recent advances in freeform surfaces, thin-plate optics, and metasurfaces provide innovative approaches to address human perception of the environment, offering promising ideas beyond conventional optical imaging. In this review, we begin with introducing the basic principles of panoramic imaging systems, and then describe the architectures, features, and functions of various panoramic imaging systems. Afterwards, we discuss in detail the broad application prospects and great design potential of freeform surfaces, thin-plate optics, and metasurfaces in panoramic imaging. We then provide a detailed analysis on how these techniques can help enhance the performance of panoramic imaging systems. We further offer a detailed analysis of applications of panoramic imaging in scene understanding for autonomous driving and robotics, spanning panoramic semantic image segmentation, panoramic depth estimation, panoramic visual localization, and so on. Finally, we cast a perspective on future potential and research directions for panoramic imaging instruments.
10.NMR: Neural Manifold Representation for Autonomous Driving ⬇️
Autonomous driving requires efficient reasoning about the Spatio-temporal nature of the semantics of the scene. Recent approaches have successfully amalgamated the traditional modular architecture of an autonomous driving stack comprising perception, prediction, and planning in an end-to-end trainable system. Such a system calls for a shared latent space embedding with interpretable intermediate trainable projected representation. One such successfully deployed representation is the Bird's-Eye View(BEV) representation of the scene in ego-frame. However, a fundamental assumption for an undistorted BEV is the local coplanarity of the world around the ego-vehicle. This assumption is highly restrictive, as roads, in general, do have gradients. The resulting distortions make path planning inefficient and incorrect. To overcome this limitation, we propose Neural Manifold Representation (NMR), a representation for the task of autonomous driving that learns to infer semantics and predict way-points on a manifold over a finite horizon, centered on the ego-vehicle. We do this using an iterative attention mechanism applied on a latent high dimensional embedding of surround monocular images and partial ego-vehicle state. This representation helps generate motion and behavior plans consistent with and cognizant of the surface geometry. We propose a sampling algorithm based on edge-adaptive coverage loss of BEV occupancy grid and associated guidance flow field to generate the surface manifold while incurring minimal computational overhead. We aim to test the efficacy of our approach on CARLA and SYNTHIA-SF.
11.An Empirical Study Of Self-supervised Learning Approaches For Object Detection With Transformers ⬇️
Self-supervised learning (SSL) methods such as masked language modeling have shown massive performance gains by pretraining transformer models for a variety of natural language processing tasks. The follow-up research adapted similar methods like masked image modeling in vision transformer and demonstrated improvements in the image classification task. Such simple self-supervised methods are not exhaustively studied for object detection transformers (DETR, Deformable DETR) as their transformer encoder modules take input in the convolutional neural network (CNN) extracted feature space rather than the image space as in general vision transformers. However, the CNN feature maps still maintain the spatial relationship and we utilize this property to design self-supervised learning approaches to train the encoder of object detection transformers in pretraining and multi-task learning settings. We explore common self-supervised methods based on image reconstruction, masked image modeling and jigsaw. Preliminary experiments in the iSAID dataset demonstrate faster convergence of DETR in the initial epochs in both pretraining and multi-task learning settings; nonetheless, similar improvement is not observed in the case of multi-task learning with Deformable DETR. The code for our experiments with DETR and Deformable DETR are available at this https URL and this https URL respectively.
12.READ: Large-Scale Neural Scene Rendering for Autonomous Driving ⬇️
Synthesizing free-view photo-realistic images is an important task in multimedia. With the development of advanced driver assistance systems~(ADAS) and their applications in autonomous vehicles, experimenting with different scenarios becomes a challenge. Although the photo-realistic street scenes can be synthesized by image-to-image translation methods, which cannot produce coherent scenes due to the lack of 3D information. In this paper, a large-scale neural rendering method is proposed to synthesize the autonomous driving scene~(READ), which makes it possible to synthesize large-scale driving scenarios on a PC through a variety of sampling schemes. In order to represent driving scenarios, we propose an {\omega} rendering network to learn neural descriptors from sparse point clouds. Our model can not only synthesize realistic driving scenes but also stitch and edit driving scenes. Experiments show that our model performs well in large-scale driving scenarios.
13.TextMatcher: Cross-Attentional Neural Network to Compare Image and Text ⬇️
We study a novel multimodal-learning problem, which we call text matching: given an image containing a single-line text and a candidate text transcription, the goal is to assess whether the text represented in the image corresponds to the candidate text. We devise the first machine-learning model specifically designed for this problem. The proposed model, termed TextMatcher, compares the two inputs by applying a cross-attention mechanism over the embedding representations of image and text, and it is trained in an end-to-end fashion. We extensively evaluate the empirical performance of TextMatcher on the popular IAM dataset. Results attest that, compared to a baseline and existing models designed for related problems, TextMatcher achieves higher performance on a variety of configurations, while at the same time running faster at inference time. We also showcase TextMatcher in a real-world application scenario concerning the automatic processing of bank cheques.
14.Deep Learning and Computer Vision Techniques for Microcirculation Analysis: A Review ⬇️
The analysis of microcirculation images has the potential to reveal early signs of life-threatening diseases like sepsis. Quantifying the capillary density and the capillary distribution in microcirculation images can be used as a biological marker to assist critically ill patients. The quantification of these biological markers is labor-intensive, time-consuming, and subject to interobserver variability. Several computer vision techniques with varying performance can be used to automate the analysis of these microcirculation images in light of the stated challenges. In this paper, we present a survey of over 50 research papers and present the most relevant and promising computer vision algorithms to automate the analysis of microcirculation images. Furthermore, we present a survey of the methods currently used by other researchers to automate the analysis of microcirculation images. This survey is of high clinical relevance because it acts as a guidebook of techniques for other researchers to develop their microcirculation analysis systems and algorithms.
15.Scene Consistency Representation Learning for Video Scene Segmentation ⬇️
A long-term video, such as a movie or TV show, is composed of various scenes, each of which represents a series of shots sharing the same semantic story. Spotting the correct scene boundary from the long-term video is a challenging task, since a model must understand the storyline of the video to figure out where a scene starts and ends. To this end, we propose an effective Self-Supervised Learning (SSL) framework to learn better shot representations from unlabeled long-term videos. More specifically, we present an SSL scheme to achieve scene consistency, while exploring considerable data augmentation and shuffling methods to boost the model generalizability. Instead of explicitly learning the scene boundary features as in the previous methods, we introduce a vanilla temporal model with less inductive bias to verify the quality of the shot features. Our method achieves the state-of-the-art performance on the task of Video Scene Segmentation. Additionally, we suggest a more fair and reasonable benchmark to evaluate the performance of Video Scene Segmentation methods. The code is made available.
16.Contrastive Supervised Distillation for Continual Representation Learning ⬇️
In this paper, we propose a novel training procedure for the continual representation learning problem in which a neural network model is sequentially learned to alleviate catastrophic forgetting in visual search tasks. Our method, called Contrastive Supervised Distillation (CSD), reduces feature forgetting while learning discriminative features. This is achieved by leveraging labels information in a distillation setting in which the student model is contrastively learned from the teacher model. Extensive experiments show that CSD performs favorably in mitigating catastrophic forgetting by outperforming current state-of-the-art methods. Our results also provide further evidence that feature forgetting evaluated in visual retrieval tasks is not as catastrophic as in classification tasks. Code at: this https URL.
17.A Continual Deepfake Detection Benchmark: Dataset, Methods, and Essentials ⬇️
There have been emerging a number of benchmarks and techniques for the detection of deepfakes. However, very few works study the detection of incrementally appearing deepfakes in the real-world scenarios. To simulate the wild scenes, this paper suggests a continual deepfake detection benchmark (CDDB) over a new collection of deepfakes from both known and unknown generative models. The suggested CDDB designs multiple evaluations on the detection over easy, hard, and long sequence of deepfake tasks, with a set of appropriate measures. In addition, we exploit multiple approaches to adapt multiclass incremental learning methods, commonly used in the continual visual recognition, to the continual deepfake detection problem. We evaluate several methods, including the adapted ones, on the proposed CDDB. Within the proposed benchmark, we explore some commonly known essentials of standard continual learning. Our study provides new insights on these essentials in the context of continual deepfake detection. The suggested CDDB is clearly more challenging than the existing benchmarks, which thus offers a suitable evaluation avenue to the future research. Our benchmark dataset and the source code will be made publicly available.
18.RustSEG -- Automated segmentation of corrosion using deep learning ⬇️
The inspection of infrastructure for corrosion remains a task that is typically performed manually by qualified engineers or inspectors. This task of inspection is laborious, slow, and often requires complex access. Recently, deep learning based algorithms have revealed promise and performance in the automatic detection of corrosion. However, to date, research regarding the segmentation of images for automated corrosion detection has been limited, due to the lack of availability of per-pixel labelled data sets which are required for model training. Herein, a novel deep learning approach (termed RustSEG) is presented, that can accurately segment images for automated corrosion detection, without the requirement of per-pixel labelled data sets for training. The RustSEG method will first, using deep learning techniques, determine if corrosion is present in an image (i.e. a classification task), and then if corrosion is present, the model will examine what pixels in the original image contributed to that classification decision. Finally, the method can refine its predictions into a pixel-level segmentation mask. In ideal cases, the method is able to generate precise masks of corrosion in images, demonstrating that the automated segmentation of corrosion without per-pixel training data is possible, addressing a significant hurdle in automated infrastructure inspection.
19.Multi-Label Logo Recognition and Retrieval based on Weighted Fusion of Neural Features ⬇️
Logo classification is a particular case of image classification, since these may contain only text, images, or a combination of both. In this work, we propose a system for the multi-label classification and similarity search of logo images. The method allows obtaining the most similar logos on the basis of their shape, color, business sector, semantics, general characteristics, or a combination of such features established by the user. This is done by employing a set of multi-label networks specialized in certain characteristics of logos. The features extracted from these networks are combined to perform the similarity search according to the search criteria established. Since the text of logos is sometimes irrelevant for the classification, a preprocessing stage is carried out to remove it, thus improving the overall performance. The proposed approach is evaluated using the European Union Trademark (EUTM) dataset, structured with the hierarchical Vienna classification system, which includes a series of metadata with which to index trademarks. We also make a comparison between well known logo topologies and Vienna in order to help designers understand their correspondences. The experimentation carried out attained reliable performance results, both quantitatively and qualitatively, which outperformed the state-of-the-art results. In addition, since the semantics and classification of brands can often be subjective, we also surveyed graphic design students and professionals in order to assess the reliability of the proposed method.
20.An Objective Method for Pedestrian Occlusion Level Classification ⬇️
Pedestrian detection is among the most safety-critical features of driver assistance systems for autonomous vehicles. One of the most complex detection challenges is that of partial occlusion, where a target object is only partially available to the sensor due to obstruction by another foreground object. A number of current pedestrian detection benchmarks provide annotation for partial occlusion to assess algorithm performance in these scenarios, however each benchmark varies greatly in their definition of the occurrence and severity of occlusion. In addition, current occlusion level annotation methods contain a high degree of subjectivity by the human annotator. This can lead to inaccurate or inconsistent reporting of an algorithm's detection performance for partially occluded pedestrians, depending on which benchmark is used. This research presents a novel, objective method for pedestrian occlusion level classification for ground truth annotation. Occlusion level classification is achieved through the identification of visible pedestrian keypoints and through the use of a novel, effective method of 2D body surface area estimation. Experimental results demonstrate that the proposed method reflects the pixel-wise occlusion level of pedestrians in images and is effective for all forms of occlusion, including challenging edge cases such as self-occlusion, truncation and inter-occluding pedestrians.
21.Recurrent Encoder-Decoder Networks for Vessel Trajectory Prediction with Uncertainty Estimation ⬇️
Recent deep learning methods for vessel trajectory prediction are able to learn complex maritime patterns from historical Automatic Identification System (AIS) data and accurately predict sequences of future vessel positions with a prediction horizon of several hours. However, in maritime surveillance applications, reliably quantifying the prediction uncertainty can be as important as obtaining high accuracy. This paper extends deep learning frameworks for trajectory prediction tasks by exploring how recurrent encoder-decoder neural networks can be tasked not only to predict but also to yield a corresponding prediction uncertainty via Bayesian modeling of epistemic and aleatoric uncertainties. We compare the prediction performance of two different models based on labeled or unlabeled input data to highlight how uncertainty quantification and accuracy can be improved by using, if available, additional information on the intention of the ship (e.g., its planned destination).
22.AutoLC: Search Lightweight and Top-Performing Architecture for Remote Sensing Image Land-Cover Classification ⬇️
Land-cover classification has long been a hot and difficult challenge in remote sensing community. With massive High-resolution Remote Sensing (HRS) images available, manually and automatically designed Convolutional Neural Networks (CNNs) have already shown their great latent capacity on HRS land-cover classification in recent years. Especially, the former can achieve better performance while the latter is able to generate lightweight architecture. Unfortunately, they both have shortcomings. On the one hand, because manual CNNs are almost proposed for natural image processing, it becomes very redundant and inefficient to process HRS images. On the other hand, nascent Neural Architecture Search (NAS) techniques for dense prediction tasks are mainly based on encoder-decoder architecture, and just focus on the automatic design of the encoder, which makes it still difficult to recover the refined mapping when confronting complicated HRS scenes.
To overcome their defects and tackle the HRS land-cover classification problems better, we propose AutoLC which combines the advantages of two methods. First, we devise a hierarchical search space and gain the lightweight encoder underlying gradient-based search strategy. Second, we meticulously design a lightweight but top-performing decoder that is adaptive to the searched encoder of itself. Finally, experimental results on the LoveDA land-cover dataset demonstrate that our AutoLC method outperforms the state-of-art manual and automatic methods with much less computational consumption.
23.Deep Depth Completion: A Survey ⬇️
Depth completion aims at predicting dense pixel-wise depth from a sparse map captured from a depth sensor. It plays an essential role in various applications such as autonomous driving, 3D reconstruction, augmented reality, and robot navigation. Recent successes on the task have been demonstrated and dominated by deep learning based solutions. In this article, for the first time, we provide a comprehensive literature review that helps readers better grasp the research trends and clearly understand the current advances. We investigate the related studies from the design aspects of network architectures, loss functions, benchmark datasets, and learning strategies with a proposal of a novel taxonomy that categorizes existing methods. Besides, we present a quantitative comparison of model performance on two widely used benchmark datasets, including an indoor and an outdoor dataset. Finally, we discuss the challenges of prior works and provide readers with some insights for future research directions.
24.Arbitrary Shape Text Detection via Boundary Transformer ⬇️
Arbitrary shape text detection is a challenging task due to its complexity and variety, e.g, various scales, random rotations, and curve shapes. In this paper, we propose an arbitrary shape text detector with a boundary transformer, which can accurately and directly locate text boundaries without any post-processing. Our method mainly consists of a boundary proposal module and an iteratively optimized boundary transformer module. The boundary proposal module consisting of multi-layer dilated convolutions will compute important prior information (including classification map, distance field, and direction field) for generating coarse boundary proposals meanwhile guiding the optimization of boundary transformer. The boundary transformer module adopts an encoder-decoder structure, in which the encoder is constructed by multi-layer transformer blocks with residual connection while the decoder is a simple multi-layer perceptron network (MLP). Under the guidance of prior information, the boundary transformer module will gradually refine the coarse boundary proposals via boundary deformation in an iterative manner. Furthermore, we propose a novel boundary energy loss (BEL) which introduces an energy minimization constraint and an energy monotonically decreasing constraint for every boundary optimization step. Extensive experiments on publicly available and challenging datasets demonstrate the state-of-the-art performance and promising efficiency of our method.
25.Invisible-to-Visible: Privacy-Aware Human Segmentation using Airborne Ultrasound via Collaborative Learning Probabilistic U-Net ⬇️
Color images are easy to understand visually and can acquire a great deal of information, such as color and texture. They are highly and widely used in tasks such as segmentation. On the other hand, in indoor person segmentation, it is necessary to collect person data considering privacy. We propose a new task for human segmentation from invisible information, especially airborne ultrasound. We first convert ultrasound waves to reflected ultrasound directional images (ultrasound images) to perform segmentation from invisible information. Although ultrasound images can roughly identify a person's location, the detailed shape is ambiguous. To address this problem, we propose a collaborative learning probabilistic U-Net that uses ultrasound and segmentation images simultaneously during training, closing the probabilistic distributions between ultrasound and segmentation images by comparing the parameters of the latent spaces. In inference, only ultrasound images can be used to obtain segmentation results. As a result of performance verification, the proposed method could estimate human segmentations more accurately than conventional probabilistic U-Net and other variational autoencoder models.
26.ReFine: Re-randomization before Fine-tuning for Cross-domain Few-shot Learning ⬇️
Cross-domain few-shot learning (CD-FSL), where there are few target samples under extreme differences between source and target domains, has recently attracted huge attention. For CD-FSL, recent studies generally have developed transfer learning based approaches that pre-train a neural network on popular labeled source domain datasets and then transfer it to target domain data. Although the labeled datasets may provide suitable initial parameters for the target data, the domain difference between the source and target might hinder the fine-tuning on the target domain. This paper proposes a simple yet powerful method that re-randomizes the parameters fitted on the source domain before adapting to the target data. The re-randomization resets source-specific parameters of the source pre-trained model and thus facilitates fine-tuning on the target domain, improving few-shot performance.
27.AggPose: Deep Aggregation Vision Transformer for Infant Pose Estimation ⬇️
Movement and pose assessment of newborns lets experienced pediatricians predict neurodevelopmental disorders, allowing early intervention for related diseases. However, most of the newest AI approaches for human pose estimation methods focus on adults, lacking publicly benchmark for infant pose estimation. In this paper, we fill this gap by proposing infant pose dataset and Deep Aggregation Vision Transformer for human pose estimation, which introduces a fast trained full transformer framework without using convolution operations to extract features in the early stages. It generalizes Transformer + MLP to high-resolution deep layer aggregation within feature maps, thus enabling information fusion between different vision levels. We pre-train AggPose on COCO pose dataset and apply it on our newly released large-scale infant pose estimation dataset. The results show that AggPose could effectively learn the multi-scale features among different resolutions and significantly improve the performance of infant pose estimation. We show that AggPose outperforms hybrid model HRFormer and TokenPose in the infant pose estimation dataset. Moreover, our AggPose outperforms HRFormer by 0.7% AP on COCO val pose estimation on average. Our code is available at this http URL.
28.Spatial-Temporal Space Hand-in-Hand: Spatial-Temporal Video Super-Resolution via Cycle-Projected Mutual Learning ⬇️
Spatial-Temporal Video Super-Resolution (ST-VSR) aims to generate super-resolved videos with higher resolution(HR) and higher frame rate (HFR). Quite intuitively, pioneering two-stage based methods complete ST-VSR by directly combining two sub-tasks: Spatial Video Super-Resolution (S-VSR) and Temporal Video Super-Resolution(T-VSR) but ignore the reciprocal relations among them. Specifically, 1) T-VSR to S-VSR: temporal correlations help accurate spatial detail representation with more clues; 2) S-VSR to T-VSR: abundant spatial information contributes to the refinement of temporal prediction. To this end, we propose a one-stage based Cycle-projected Mutual learning network (CycMu-Net) for ST-VSR, which makes full use of spatial-temporal correlations via the mutual learning between S-VSR and T-VSR. Specifically, we propose to exploit the mutual information among them via iterative up-and-down projections, where the spatial and temporal features are fully fused and distilled, helping the high-quality video reconstruction. Besides extensive experiments on benchmark datasets, we also compare our proposed CycMu-Net with S-VSR and T-VSR tasks, demonstrating that our method significantly outperforms state-of-the-art methods.
29.Salient Object Detection via Bounding-box Supervision ⬇️
The success of fully supervised saliency detection models depends on a large number of pixel-wise labeling. In this paper, we work on bounding-box based weakly-supervised saliency detection to relieve the labeling effort. Given the bounding box annotation, we observe that pixels inside the bounding box may contain extensive labeling noise. However, as a large amount of background is excluded, the foreground bounding box region contains a less complex background, making it possible to perform handcrafted features-based saliency detection with only the cropped foreground region. As the conventional handcrafted features are not representative enough, leading to noisy saliency maps, we further introduce structure-aware self-supervised loss to regularize the structure of the prediction. Further, we claim that pixels outside the bounding box should be background, thus partial cross-entropy loss function can be used to accurately localize the accurate background region. Experimental results on six benchmark RGB saliency datasets illustrate the effectiveness of our model.
30.DcnnGrasp: Towards Accurate Grasp Pattern Recognition with Adaptive Regularizer Learning ⬇️
The task of grasp pattern recognition aims to derive the applicable grasp types of an object according to the visual information. Current state-of-the-art methods ignore category information of objects which is crucial for grasp pattern recognition. This paper presents a novel dual-branch convolutional neural network (DcnnGrasp) to achieve joint learning of object category classification and grasp pattern recognition. DcnnGrasp takes object category classification as an auxiliary task to improve the effectiveness of grasp pattern recognition. Meanwhile, a new loss function called joint cross-entropy with an adaptive regularizer is derived through maximizing a posterior, which significantly improves the model performance. Besides, based on the new loss function, a training strategy is proposed to maximize the collaborative learning of the two tasks. The experiment was performed on five household objects datasets including the RGB-D Object dataset, Hit-GPRec dataset, Amsterdam library of object images (ALOI), Columbia University Image Library (COIL-100), and MeganePro dataset 1. The experimental results demonstrated that the proposed method can achieve competitive performance on grasp pattern recognition with several state-of-the-art methods. Specifically, our method even outperformed the second-best one by nearly 15% in terms of global accuracy for the case of testing a novel object on the RGB-D Object dataset.
31.Student Collaboration Improves Self-Supervised Learning: Dual-Loss Adaptive Masked Autoencoder for Brain Cell Image Analysis ⬇️
Self-supervised learning leverages the underlying data structure as the source of the supervisory signal without the need for human annotation effort. This approach offers a practical solution to learning with a large amount of biomedical data and limited annotation. Unlike other studies exploiting data via multi-view (e.g., augmented images), this study presents a self-supervised Dual-Loss Adaptive Masked Autoencoder (DAMA) algorithm established from the viewpoint of the information theory. Specifically, our objective function maximizes the mutual information by minimizing the conditional entropy in pixel-level reconstruction and feature-level regression. We further introduce an adaptive mask sampling strategy to maximize mutual information. We conduct extensive experiments on brain cell images to validate the proposed method. DAMA significantly outperforms both state-of-the-art self-supervised and supervised methods on brain cells data and demonstrates competitive result on ImageNet-1k. Code: this https URL
32.On Scale Space Radon Transform, Properties and Image Reconstruction ⬇️
Aware of the importance of the good behavior in the scale space that a mathematical transform must have, we depict, in this paper, the basic properties and the inverse transform of the Scale Space Radon Transform (SSRT). To reconstruct the image from SSRT sinogram, the Filtered backprojection (FBP) technique is used in two different ways: (1) Deconvolve SSRT to obtain the estimated Radon transform (RT) and then, reconstruct image using classical FBP or (2) Adapt FBP technique to SSRT so that the Radon projections spectrum used in classical FBP is replaced by SSRT and Wiener filtering, expressed in the frequency domain. Comparison of image reconstruction techniques using SSRT and RT are performed on Shepp-Logan head phantom image. Using the Mean Absolute Error (MAE) as image reconstruction quality measure, the preliminary results present an outstanding performance for SSRT-based image reconstruction techniques compared to the RT-based one. Furthermore, the method (2) outperforms the method (1) in terms of computation time and adaptability for high level of noise when fairly large Gaussian kernel is used.
33.Robustness of Humans and Machines on Object Recognition with Extreme Image Transformations ⬇️
Recent neural network architectures have claimed to explain data from the human visual cortex. Their demonstrated performance is however still limited by the dependence on exploiting low-level features for solving visual tasks. This strategy limits their performance in case of out-of-distribution/adversarial data. Humans, meanwhile learn abstract concepts and are mostly unaffected by even extreme image distortions. Humans and networks employ strikingly different strategies to solve visual tasks. To probe this, we introduce a novel set of image transforms and evaluate humans and networks on an object recognition task. We found performance for a few common networks quickly decreases while humans are able to recognize objects with a high accuracy.
34.Few-Shot Image Classification Benchmarks are Too Far From Reality: Build Back Better with Semantic Task Sampling ⬇️
Every day, a new method is published to tackle Few-Shot Image Classification, showing better and better performances on academic benchmarks. Nevertheless, we observe that these current benchmarks do not accurately represent the real industrial use cases that we encountered. In this work, through both qualitative and quantitative studies, we expose that the widely used benchmark tieredImageNet is strongly biased towards tasks composed of very semantically dissimilar classes e.g. bathtub, cabbage, pizza, schipperke, and cardoon. This makes tieredImageNet (and similar benchmarks) irrelevant to evaluate the ability of a model to solve real-life use cases usually involving more fine-grained classification. We mitigate this bias using semantic information about the classes of tieredImageNet and generate an improved, balanced benchmark. Going further, we also introduce a new benchmark for Few-Shot Image Classification using the Danish Fungi 2020 dataset. This benchmark proposes a wide variety of evaluation tasks with various fine-graininess. Moreover, this benchmark includes many-way tasks (e.g. composed of 100 classes), which is a challenging setting yet very common in industrial applications. Our experiments bring out the correlation between the difficulty of a task and the semantic similarity between its classes, as well as a heavy performance drop of state-of-the-art methods on many-way few-shot classification, raising questions about the scaling abilities of these methods. We hope that our work will encourage the community to further question the quality of standard evaluation processes and their relevance to real-life applications.
35.End-to-End Multi-Person Audio/Visual Automatic Speech Recognition ⬇️
Traditionally, audio-visual automatic speech recognition has been studied under the assumption that the speaking face on the visual signal is the face matching the audio. However, in a more realistic setting, when multiple faces are potentially on screen one needs to decide which face to feed to the A/V ASR system. The present work takes the recent progress of A/V ASR one step further and considers the scenario where multiple people are simultaneously on screen (multi-person A/V ASR). We propose a fully differentiable A/V ASR model that is able to handle multiple face tracks in a video. Instead of relying on two separate models for speaker face selection and audio-visual ASR on a single face track, we introduce an attention layer to the ASR encoder that is able to soft-select the appropriate face video track. Experiments carried out on an A/V system trained on over 30k hours of YouTube videos illustrate that the proposed approach can automatically select the proper face tracks with minor WER degradation compared to an oracle selection of the speaking face while still showing benefits of employing the visual signal instead of the audio alone.
36.DoubleMatch: Improving Semi-Supervised Learning with Self-Supervision ⬇️
Following the success of supervised learning, semi-supervised learning (SSL) is now becoming increasingly popular. SSL is a family of methods, which in addition to a labeled training set, also use a sizable collection of unlabeled data for fitting a model. Most of the recent successful SSL methods are based on pseudo-labeling approaches: letting confident model predictions act as training labels. While these methods have shown impressive results on many benchmark datasets, a drawback of this approach is that not all unlabeled data are used during training. We propose a new SSL algorithm, DoubleMatch, which combines the pseudo-labeling technique with a self-supervised loss, enabling the model to utilize all unlabeled data in the training process. We show that this method achieves state-of-the-art accuracies on multiple benchmark datasets while also reducing training times compared to existing SSL methods. Code is available at this https URL.
37.Performance of a deep learning system for detection of referable diabetic retinopathy in real clinical settings ⬇️
Background: To determine the ability of a commercially available deep learning system, RetCAD v.1.3.1 (Thirona, Nijmegen, The Netherlands) for the automatic detection of referable diabetic retinopathy (DR) on a dataset of colour fundus images acquired during routine clinical practice in a tertiary hospital screening program, analyzing the reduction of workload that can be released incorporating this artificial intelligence-based technology. Methods: Evaluation of the software was performed on a dataset of 7195 nonmydriatic fundus images from 6325 eyes of 3189 diabetic patients attending our screening program between February to December of 2019. The software generated a DR severity score for each colour fundus image which was combined into an eye-level score. This score was then compared with a reference standard as set by a human expert using receiver operating characteristic (ROC) curve analysis. Results: The artificial intelligence (AI) software achieved an area under the ROC curve (AUC) value of 0.988 [0.981:0.993] for the detection of referable DR. At the proposed operating point, the sensitivity of the RetCAD software for DR is 90.53% and specificity is 97.13%. A workload reduction of 96% could be achieved at the cost of only 6 false negatives. Conclusions: The AI software correctly identified the vast majority of referable DR cases, with a workload reduction of 96% of the cases that would need to be checked, while missing almost no true cases, so it may therefore be used as an instrument for triage.
38.CNN-LSTM Based Multimodal MRI and Clinical Data Fusion for Predicting Functional Outcome in Stroke Patients ⬇️
Clinical outcome prediction plays an important role in stroke patient management. From a machine learning point-of-view, one of the main challenges is dealing with heterogeneous data at patient admission, i.e. the image data which are multidimensional and the clinical data which are scalars. In this paper, a multimodal convolutional neural network - long short-term memory (CNN-LSTM) based ensemble model is proposed. For each MR image module, a dedicated network provides preliminary prediction of the clinical outcome using the modified Rankin scale (mRS). The final mRS score is obtained by merging the preliminary probabilities of each module dedicated to a specific type of MR image weighted by the clinical metadata, here age or the National Institutes of Health Stroke Scale (NIHSS). The experimental results demonstrate that the proposed model surpasses the baselines and offers an original way to automatically encode the spatio-temporal context of MR images in a deep learning architecture. The highest AUC (0.77) was achieved for the proposed model with NIHSS.
39.Secure Federated Learning for Neuroimaging ⬇️
The amount of biomedical data continues to grow rapidly. However, the ability to collect data from multiple sites for joint analysis remains challenging due to security, privacy, and regulatory concerns. We present a Secure Federated Learning architecture, MetisFL, which enables distributed training of neural networks over multiple data sources without sharing data. Each site trains the neural network over its private data for some time, then shares the neural network parameters (i.e., weights, gradients) with a Federation Controller, which in turn aggregates the local models, sends the resulting community model back to each site, and the process repeats. Our architecture provides strong security and privacy. First, sample data never leaves a site. Second, neural parameters are encrypted before transmission and the community model is computed under fully-homomorphic encryption. Finally, we use information-theoretic methods to limit information leakage from the neural model to prevent a curious site from performing membership attacks. We demonstrate this architecture in neuroimaging. Specifically, we investigate training neural models to classify Alzheimer's disease, and estimate Brain Age, from magnetic resonance imaging datasets distributed across multiple sites, including heterogeneous environments where sites have different amounts of data, statistical distributions, and computational capabilities.
40.Best of Both Worlds: Multi-task Audio-Visual Automatic Speech Recognition and Active Speaker Detection ⬇️
Under noisy conditions, automatic speech recognition (ASR) can greatly benefit from the addition of visual signals coming from a video of the speaker's face. However, when multiple candidate speakers are visible this traditionally requires solving a separate problem, namely active speaker detection (ASD), which entails selecting at each moment in time which of the visible faces corresponds to the audio. Recent work has shown that we can solve both problems simultaneously by employing an attention mechanism over the competing video tracks of the speakers' faces, at the cost of sacrificing some accuracy on active speaker detection. This work closes this gap in active speaker detection accuracy by presenting a single model that can be jointly trained with a multi-task loss. By combining the two tasks during training we reduce the ASD classification accuracy by approximately 25%, while simultaneously improving the ASR performance when compared to the multi-person baseline trained exclusively for ASR.
41.Deep fusion of gray level co-occurrence matrices for lung nodule classification ⬇️
Lung cancer is a severe menace to human health, due to which millions of people die because of late diagnoses of cancer; thus, it is vital to detect the disease as early as possible. The Computerized chest analysis Tomography of scan is assumed to be one of the efficient solutions for detecting and classifying lung nodules. The necessity of high accuracy of analyzing C.T. scan images of the lung is considered as one of the crucial challenges in detecting and classifying lung cancer. A new long-short-term-memory (LSTM) based deep fusion structure, is introduced, where, the texture features computed from lung nodules through new volumetric grey-level-co-occurrence-matrices (GLCM) computations are applied to classify the nodules into: benign, malignant and ambiguous. An improved Otsu segmentation method combined with the water strider optimization algorithm (WSA) is proposed to detect the lung nodules. Otsu-WSA thresholding can overcome the restrictions present in previous thresholding methods. Extended experiments are run to assess this fusion structure by considering 2D-GLCM computations based 2D-slices fusion, and an approximation of this 3D-GLCM with volumetric 2.5D-GLCM computations-based LSTM fusion structure. The proposed methods are trained and assessed through the LIDC-IDRI dataset, where 94.4%, 91.6%, and 95.8% Accuracy, sensitivity, and specificity are obtained, respectively for 2D-GLCM fusion and 97.33%, 96%, and 98%, accuracy, sensitivity, and specificity, respectively, for 2.5D-GLCM fusion. The yield of the same are 98.7%, 98%, and 99%, for the 3D-GLCM fusion. The obtained results and analysis indicate that the WSA-Otsu method requires less execution time and yields a more accurate thresholding process. It is found that 3D-GLCM based LSTM outperforms its counterparts.
42.An Efficient Calculation of Quaternion Correlation of Signals and Color Images ⬇️
Over the past century, a correlation has been an essential mathematical technique utilized in engineering sciences, including practically every signal/image processing field. This paper describes an effective method of calculating the correlation function of signals and color images in quaternion algebra. We propose using the quaternions with a commutative multiplication operation and defining the corresponding correlation function in this arithmetic. The correlation between quaternion signals and images can be calculated by multiplying two quaternion DFTs of signals and images. The complexity of the correlation of color images is three times higher than in complex algebra.