YOLOv3¶
Abstract¶
We present some updates to YOLO! We made a bunch of little design changes to make it better. We also trained this new network that's pretty swell. It's a little bigger than last time but more accurate. It's still fast though, don't worry. At 320x320 YOLOv3 runs in 22 ms at 28.2 mAP, as accurate as SSD but three times faster. When we look at the old .5 IOU mAP detection metric YOLOv3 is quite good. It achieves 57.9 mAP@50 in 51 ms on a Titan X, compared to 57.5 mAP@50 in 198 ms by RetinaNet, similar performance but 3.8x faster.
Results¶
performance tested on Ascend 910(8p) with graph mode
| Name | Scale | BatchSize | ImageSize | Dataset | Box mAP (%) | Params | Recipe | Download |
|---|---|---|---|---|---|---|---|---|
| YOLOv3 | Darknet53 | 16 * 8 | 640 | MS COCO 2017 | 45.5 | 61.9M | yaml | weights |
performance tested on Ascend 910*(8p)
| Name | Scale | BatchSize | ImageSize | Dataset | Box mAP (%) | ms/step | Params | Recipe | Download |
|---|---|---|---|---|---|---|---|---|---|
| YOLOv3 | Darknet53 | 16 * 8 | 640 | MS COCO 2017 | 46.6 | 396.60 | 61.9M | yaml | weights |
Notes¶
- Box mAP: Accuracy reported on the validation set.
- We referred to a commonly used third-party YOLOv3 implementation.
Quick Start¶
Please refer to the QUICK START in MindYOLO for details.
Training¶
- Pretraining Model¶
You can get the pre-training model from here.
To convert it to a loadable ckpt file for mindyolo, please put it in the root directory then run it
python mindyolo/utils/convert_weight_darknet53.py
- Distributed Training¶
It is easy to reproduce the reported results with the pre-defined training recipe. For distributed training on multiple Ascend 910 devices, please run
# distributed training on multiple GPU/Ascend devices
msrun --worker_num=8 --local_worker_num=8 --bind_core=True --log_dir=./yolov3_log python train.py --config ./configs/yolov3/yolov3.yaml --device_target Ascend --is_parallel True
Similarly, you can train the model on multiple GPU devices with the above msrun command. Note: For more information about msrun configuration, please refer to here.
For detailed illustration of all hyper-parameters, please refer to config.py.
Note: As the global batch size (batch_size x num_devices) is an important hyper-parameter, it is recommended to keep the global batch size unchanged for reproduction or adjust the learning rate linearly to a new global batch size.
- Standalone Training¶
If you want to train or finetune the model on a smaller dataset without distributed training, please run:
# standalone training on a CPU/GPU/Ascend device
python train.py --config ./configs/yolov3/yolov3.yaml --device_target Ascend
Validation and Test¶
To validate the accuracy of the trained model, you can use test.py and parse the checkpoint path with --weight.
python test.py --config ./configs/yolov3/yolov3.yaml --device_target Ascend --weight /PATH/TO/WEIGHT.ckpt
Deployment¶
See here.
References¶
[1] Jocher Glenn. YOLOv3 release v9.1. https://github.com/ultralytics/yolov3/releases/tag/v9.1, 2021. [2] Joseph Redmon and Ali Farhadi. YOLOv3: An incremental improvement. arXiv preprint arXiv:1804.02767, 2018.