The code for the implementation of “yolov5 + circular Smooth Label”.
The results on DOTAv1.5_subsize1024_gap200_rate1.0 test-dev set are shown in the table below. (password:yolo)
Model (link) |
Size (pixels) |
TTA (multi-scale/ rotate testing) |
mAPOBB 0.5 |
Speed CPU b1 (ms) |
Speed 2080Ti b1 (ms) |
params (M) |
FLOPs @1024 (B) |
---|---|---|---|---|---|---|---|
yolov5m | 1024 | × | 73.19 | - | - | 21.6 | 50.5 |
yolov5m6 | 1024 | × | - | - | - | - | - |
yolov5m7 | 1024 | × | - | - | - | - | - |
Note:
- All the pre-trained checkpoint about yolov5 can be downloaded in this link.
- [2022/1/7] : Faster and stronger, some bugs fixed
Please refer to for installation and dataset preparation.
This repo is based on . Please see for the Oriented Detection basic usage.
I have used utility functions from other wonderful open-source projects. Espeicially thank the authors of:
- ultralytics/yolov5.
- Thinklab-SJTU/CSL_RetinaNet_Tensorflow.
- jbwang1997/OBBDetection
- CAPTAIN-WHU/DOTA_devkit
想要了解相关实现的细节和原理可以看我的知乎文章:
在使用中有任何问题,欢迎反馈给我,可以用以下联系方式跟我交流
- 知乎(@略略略)
- 代码问题提issues,其他问题请知乎上联系
Name : "胡凯旋"
describe myself:"咸鱼一枚"