mirrors
/
mmyolo
mirror of https://github.com/open-mmlab/mmyolo.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

fix yolox-l ckpt link (#677)

pull/690/head
Peng Lu 2023-03-20 16:23:15 +08:00 committed by GitHub
parent 624602c541
commit 8d67cea185
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
configs/yolox/README.md
View File

@ -27,7 +27,7 @@ YOLOX-l model structure
| YOLOX-s | 640 | 8xb32 | Yes | Yes | 9.8 | 41.9 (+1.2) | [config](./yolox_s_fast_8xb32-300e-rtmdet-hyp_coco.py) | [model](https://download.openmmlab.com/mmyolo/v0/yolox/yolox_s_fast_8xb32-300e-rtmdet-hyp_coco/yolox_s_fast_8xb32-300e-rtmdet-hyp_coco_20230210_134645-3a8dfbd7.pth) \| [log](https://download.openmmlab.com/mmyolo/v0/yolox/yolox_s_fast_8xb32-300e-rtmdet-hyp_coco/yolox_s_fast_8xb32-300e-rtmdet-hyp_coco_20230210_134645.log.json) |
| YOLOX-m | 640 | 8xb8 | Yes | No | 4.9 | 46.9 | [config](./yolox_m_fast_8xb8-300e_coco.py) | [model](https://download.openmmlab.com/mmyolo/v0/yolox/yolox_m_fast_8xb8-300e_coco/yolox_m_fast_8xb8-300e_coco_20230213_160218-a71a6b25.pth) \| [log](https://download.openmmlab.com/mmyolo/v0/yolox/yolox_m_fast_8xb8-300e_coco/yolox_m_fast_8xb8-300e_coco_20230213_160218.log.json) |
| YOLOX-m | 640 | 8xb32 | Yes | Yes | 17.6 | 47.5 (+0.6) | [config](./yolox_m_fast_8xb32-300e-rtmdet-hyp_coco.py) | [model](https://download.openmmlab.com/mmyolo/v0/yolox/yolox_m_fast_8xb32-300e-rtmdet-hyp_coco/yolox_m_fast_8xb32-300e-rtmdet-hyp_coco_20230210_144328-e657e182.pth) \| [log](https://download.openmmlab.com/mmyolo/v0/yolox/yolox_m_fast_8xb32-300e-rtmdet-hyp_coco/yolox_m_fast_8xb32-300e-rtmdet-hyp_coco_20230210_144328.log.json) |
| YOLOX-l | 640 | 8xb8 | Yes | No | 8.0 | 50.1 | [config](./yolox_l_fast_8xb8-300e_coco.py) | [model](https://download.openmmlab.com/mmyolo/v0/yolox/yolox_l_fast__8xb8-300e_coco/yolox_l_fast_8xb8-300e_coco_20230213_160715-c731eb1c.pth) \| [log](https://download.openmmlab.com/mmyolo/v0/yolox/yolox_l_fast_8xb8-300e_coco/yolox_l_fast_8xb8-300e_coco_20230213_160715.log.json) |
| YOLOX-l | 640 | 8xb8 | Yes | No | 8.0 | 50.1 | [config](./yolox_l_fast_8xb8-300e_coco.py) | [model](https://download.openmmlab.com/mmyolo/v0/yolox/yolox_l_fast_8xb8-300e_coco/yolox_l_fast_8xb8-300e_coco_20230213_160715-c731eb1c.pth) \| [log](https://download.openmmlab.com/mmyolo/v0/yolox/yolox_l_fast_8xb8-300e_coco/yolox_l_fast_8xb8-300e_coco_20230213_160715.log.json) |
| YOLOX-x | 640 | 8xb8 | Yes | No | 9.8 | 51.4 | [config](./yolox_x_fast_8xb8-300e_coco.py) | [model](https://download.openmmlab.com/mmyolo/v0/yolox/yolox_x_fast_8xb8-300e_coco/yolox_x_fast_8xb8-300e_coco_20230215_133950-1d509fab.pth) \| [log](https://download.openmmlab.com/mmyolo/v0/yolox/yolox_x_fast_8xb8-300e_coco/yolox_x_fast_8xb8-300e_coco_20230215_133950.log.json) |
YOLOX uses a default training configuration of `8xbs8` which results in a long training time, we expect it to use `8xbs32` to speed up the training and not cause a decrease in mAP. We modified `train_batch_size_per_gpu` from 8 to 32, `batch_augments_interval` from 10 to 1 and `base_lr` from 0.01 to 0.04 under YOLOX-s default configuration based on the linear scaling rule, which resulted in mAP degradation. Finally, I found that using RTMDet's training hyperparameter can improve performance in YOLOX Tiny/S/M, which also validates the superiority of RTMDet's training hyperparameter.