mirror of https://github.com/open-mmlab/mmyolo.git
27 lines
3.5 KiB
Markdown
27 lines
3.5 KiB
Markdown
|
# YOLOv5
|
||
|
|
|
||
|
|
<!-- [ALGORITHM] -->
|
||
|
|
|
||
|
|
## Abstract
|
||
|
|
|
||
|
|
YOLOv5 is a family of object detection architectures and models pretrained on the COCO dataset, and represents Ultralytics open-source research into future vision AI methods, incorporating lessons learned and best practices evolved over thousands of hours of research and development.
|
||
|
|
|
||
|
|
## Results and models
|
||
|
|
|
||
|
|
### COCO
|
||
|
|
|
||
|
|
| Backbone | size | SyncBN | AMP | Mem (GB) | box AP | Config | Download |
|
||
|
|
| :------: | :--: | :----: | :-: | :------: | :----: | :-------------------------------------------------------------------------------------------------------------------: | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
|
||
|
|
| YOLOv5-n | 640 | Yes | Yes | xxx | xxx | [config](https://github.com/open-mmlab/mmyolo/tree/master/configs/yolov5/yolov5_n-v61_syncbn_fast_8xb16-300e_coco.py) | [model](https://download.openmmlab.com/mmyolo/v0.0.1/yolov5/yolov5_n-v61_syncbn_fast_8xb16-300e_coco/) \| [log](https://download.openmmlab.com/mmyolo/v0.0.1/yolov5/yolov5_n-v61_syncbn_fast_8xb16-300e_coco/) |
|
||
|
|
| YOLOv5-s | 640 | Yes | Yes | xxx | xxx | [config](https://github.com/open-mmlab/mmyolo/tree/master/configs/yolov5/yolov5_s-v61_syncbn_fast_8xb16-300e_coco.py) | [model](https://download.openmmlab.com/mmyolo/v0.0.1/yolov5/yolov5_s-v61_syncbn_fast_8xb16-300e_coco/) \| [log](https://download.openmmlab.com/mmyolo/v0.0.1/yolov5/yolov5_s-v61_syncbn_fast_8xb16-300e_coco/) |
|
||
|
|
| YOLOv5-m | 640 | Yes | Yes | xxx | xxx | [config](https://github.com/open-mmlab/mmyolo/tree/master/configs/yolov5/yolov5_m-v61_syncbn_fast_8xb16-300e_coco.py) | [model](https://download.openmmlab.com/mmyolo/v0.0.1/yolov5/yolov5_m-v61_syncbn_fast_8xb16-300e_coco/) \| [log](https://download.openmmlab.com/mmyolo/v0.0.1/yolov5/yolov5_m-v61_syncbn_fast_8xb16-300e_coco/) |
|
||
|
|
| YOLOv5-l | 640 | Yes | Yes | xxx | xxx | [config](https://github.com/open-mmlab/mmyolo/tree/master/configs/yolov5/yolov5_l-v61_syncbn_fast_8xb16-300e_coco.py) | [model](https://download.openmmlab.com/mmyolo/v0.0.1/yolov5/yolov5_l-v61_syncbn_fast_8xb16-300e_coco/) \| [log](https://download.openmmlab.com/mmyolo/v0.0.1/yolov5/yolov5_l-v61_syncbn_fast_8xb16-300e_coco/) |
|
||
|
|
| YOLOv5-x | 640 | Yes | Yes | xxx | xxx | [config](https://github.com/open-mmlab/mmyolo/tree/master/configs/yolov5/yolov5_x-v61_syncbn_fast_8xb16-300e_coco.py) | [model](https://download.openmmlab.com/mmyolo/v0.0.1/yolov5/yolov5_x-v61_syncbn_fast_8xb16-300e_coco/) \| [log](https://download.openmmlab.com/mmyolo/v0.0.1/yolov5/yolov5_x-v61_syncbn_fast_8xb16-300e_coco/) |
|
||
|
|
|
||
|
|
**Note**:
|
||
|
|
|
||
|
|
1. `fast` means that `YOLOv5DetDataPreprocessor` and `yolov5_collate` are used for data preprocessing, which is faster for training, but less flexible for multitasking. Recommended to use fast version config if you only care about object detection.
|
||
|
|
2. `SyncBN` means use SyncBN, `AMP` indicates training with mixed precision.
|
||
|
|
3. We use 8x A100 for training, and the single-GPU batch size is 16. This is different from the official code.
|
||
|
|
4. The performance is unstable and may fluctuate by about 0.4 mAP. mAP 37.3 ~ 37.7 is acceptable in `YOLOv5-s`.
|