DynamicRCNN-mmdet

Dynamic R-CNN: Towards High Quality Object Detection via Dynamic Training

paper address: https://arxiv.org/pdf/2004.06002.pdf

Requirements

• Linux OS
• Python 3.7 (Python 2 is not supported)
• PyTorch 1.2.0
• torchvision 0.4.0
• mmdetection tag: v1.2.0
• mmcv 0.4.3
• CUDA 10.0
• GCC(G++) 5.4.0 or higher

Installation

a. Create a conda virtual environment and activate it (Optional but recommended).

conda create --name dynamic python=3.7
conda activate dynamic

b. Install pytorch and torchvision.
pip is recommended,

pip install torch==1.2.0 torchvision==0.4.0  

c. Install mmdet (other dependencies wil be installed automatically).

pip install cython
pip install easydict
pip install -r requirements.txt
pip install -v -e .

d. Prepare dataset and checkpoint file.

Download coco dataset and checkpoint file

Fold structure should be as follows:

DynamicRCNN-mmdet
├── data
│   ├── coco
│   │   ├── annotations
│   │   ├── train2017
│   │   ├── val2017
│   │   ├── test2017
├── backbone
│   ├── resnet50-19c8e357.pth

Features

• Synchronize on multi-gpus

Train

bash scripts/train.sh

Test

bash scripts/test.sh

Results

Faster-RCNN