TOV_mmdetection
TOV_mmdetection copied to clipboard
Include mmdetection version of TinyBenchmark. Official link.
trafficstars
TOV mmdetection
TOV mmdetection is an open source toolbox for object localization and detection tasks on top of mmdetection.
- Scale Match for TinyPerson Detection (WACV2020)
- Object Localization under Single Coarse Point Supervision (CVPR2022)
Introduction
TODO list:
- [x] add TinyPerson dataset and evaluation
- [x] add crop and merge for image during inference
- [x] implement RetinaNet and Faster-FPN baseline on TinyPerson
- [x] add SM/MSM experiment support
Install
install environment
conda create -n open-mmlab python=3.7 -y
conda activate open-mmlab
# install latest pytorch prebuilt with the default prebuilt CUDA version (usually the latest)
conda install -c pytorch pytorch torchvision -y
# conda install -c pytorch pytorch=1.5.0 cudatoolkit=10.2 torchvision -y
# install the latest mmcv
pip install mmcv-full --user
conda install scikit-image
download and project setting
- [note]: if your need to modified from origin mmdetection code, see here, otherwise do not need any other modified.
- [note]: for more about evaluation, see evaluation_of_tiny_object.md
git clone https://github.com/ucas-vg/TOV_mmdetection # from github
# git clone https://gitee.com/ucas-vg/TOV_mmdetection # from gitee
cd TOV_mmdetection
# download code for evaluation
git clone https://github.com/yinglang/huicv/ # from github
# git clone https://gitee.com/ucas-vg/huicv # from gitee
# install mmdetection
pip uninstall pycocotools # sometimes need to source deactivate before, for
pip install -r requirements/build.txt
pip install -v -e . --user # or try "python setup.py develop" if get still got pycocotools error
Citation
If you use the code and benchmark in your research, please cite:
@inproceedings{yu2020scale,
title={Scale Match for Tiny Person Detection},
author={Yu, Xuehui and Gong, Yuqi and Jiang, Nan and Ye, Qixiang and Han, Zhenjun},
booktitle={The IEEE Winter Conference on Applications of Computer Vision},
pages={1257--1265},
year={2020}
}
And if the ECCVW challenge sumarry do some help for your research, please cite:
@article{yu20201st,
title={The 1st Tiny Object Detection Challenge: Methods and Results},
author={Yu, Xuehui and Han, Zhenjun and Gong, Yuqi and Jan, Nan and Zhao, Jian and Ye, Qixiang and Chen, Jie and Feng, Yuan and Zhang, Bin and Wang, Xiaodi and others},
journal={arXiv preprint arXiv:2009.07506},
year={2020}
}