Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/megvii-model/YOLOF
https://github.com/megvii-model/YOLOF
Last synced: 3 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/megvii-model/YOLOF
- Owner: megvii-model
- License: mit
- Created: 2021-03-11T06:42:12.000Z (over 3 years ago)
- Default Branch: main
- Last Pushed: 2021-04-12T15:14:17.000Z (over 3 years ago)
- Last Synced: 2024-04-20T05:34:52.461Z (7 months ago)
- Language: Python
- Size: 793 KB
- Stars: 806
- Watchers: 20
- Forks: 115
- Open Issues: 27
-
Metadata Files:
- Readme: README.md
- Contributing: CONTRIBUTING.md
- License: LICENSE
Awesome Lists containing this project
- StarryDivineSky - megvii-model/YOLOF
- awesome-yolo-object-detection - YOLOF - model/YOLOF?style=social"/> : "You Only Look One-level Feature". (**[CVPR 2021](https://openaccess.thecvf.com/content/CVPR2021/html/Chen_You_Only_Look_One-Level_Feature_CVPR_2021_paper.html)**). "微信公众号「计算机视觉研究院」《[CVPR目标检测新框架：不再是YOLO，而是只需要一层特征（干货满满，建议收藏）](https://mp.weixin.qq.com/s/5sTxdjhKIPpQ-rCsWfe80A)》"。 (Summary)
README
# You Only Look One-level Feature (YOLOF), CVPR2021
A simple, fast, and efficient object detector **without** FPN.
- This repo provides an implementation for YOLOF based on [cvpods](https://github.com/Megvii-BaseDetection/cvpods). A neat and re-organized
Detectron2 version of YOLOF is available at [https://github.
com/chensnathan/YOLOF](https://github.com/chensnathan/YOLOF).
> [**You Only Look One-level Feature**](https://arxiv.org/abs/2103.09460),
> Qiang Chen, Yingming Wang, Tong Yang, Xiangyu Zhang, Jian Cheng, Jian Sun
## Getting Started
- Install `cvpods`
```shell
cd YOLOF/
python setup.py develop
```
- Install `mish-cuda` to speed up the training and inference when using `CSPDarkNet-53` as the backbone (**optional**)
```shell
git clone https://github.com/thomasbrandon/mish-cuda
cd mish-cuda
python setup.py build install
cd ..
```
- Download the pretrained model in [OneDrive](https://1drv.ms/u/s!AgM0VtBH3kV9imGxZX3n_TMQGtbP?e=YMgpGJ) or in the [Baidu Cloud](https://pan.baidu.com/s/1BSOncRYq6HeCQ8q2hrWowA) with code `qr6o` to train with the **CSPDarkNet-53** backbone (**optional**)
```shell
mkdir pretrained_models
# download the `cspdarknet53.pth` to the `pretrained_models` directory
```
- Train
```shell
cd playground/detection/coco/yolof/yolof.res50.C5.1x
pods_train --num-gpus 8
```
- Test
```shell
cd playground/detection/coco/yolof/yolof.res50.C5.1x
pods_test --num-gpus 8 MODEL.WEIGHTS /path/to/checkpoint_file
```
## Main results
The models listed below can be found in [this onedrive link](https://1drv.ms/u/s!AgM0VtBH3kV9imGxZX3n_TMQGtbP?e=YMgpGJ) or in the [BaiduCloud link](https://pan.baidu.com/s/1BSOncRYq6HeCQ8q2hrWowA) with code `qr6o`. The FPS is tested on a 2080Ti GPU.
More models will be available in the near future.
| Model | COCO val mAP | FPS |
|-------------------------------------------|---------------|-------|
| YOLOF_R_50_C5_1x | 37.7 | 32 |
| YOLOF_R_50_DC5_1x | 39.2 | 20 |
| YOLOF_R_101_C5_1x | 39.8 | 21 |
| YOLOF_R_101_DC5_1x | 40.5 | 15 |
| YOLOF_X_101_64x4d_C5_1x | 42.2 | 10 |
| YOLOF_CSP_D_53_DC5_3x | 41.2 | 39 |
| YOLOF_CSP_D_53_DC5_9x | 42.8 | 39 |
| YOLOF_CSP_D_53_DC5_9x_stage2_3x | 43.2 | 39 |
## Citation
If you find this project useful for your research, please use the following BibTeX entry.
@inproceedings{chen2021you,
title={You Only Look One-level Feature},
author={Chen, Qiang and Wang, Yingming and Yang, Tong and Zhang, Xiangyu and Cheng, Jian and Sun, Jian},
booktitle={IEEE Conference on Computer Vision and Pattern Recognition},
year={2021}
}