YOLO_Series
YOLOv3, YOLOv4, YOLOv5, YOLOX的PyTorch实现。(持续更新中......)
Quick start
- Make sure that you have installed PyTorch 1.10.0 and torchvision 0.11.1 or higher.
- Run the following to install dependencies.
pip install -r requirements.txt
- Install
pycocotools
. - Download COCO2017 and VOC2012, and then extract them under
data
folder, make them look like this:
|-- data
|-- coco
| |-- annotations
| | |-- instances_train2017.json
| | `-- instances_val2017.json
| `-- images
| |-- train2017
| | |-- ...
| `-- val2017
| |-- ...
|-- VOCdevkit
`-- |-- Annotations
|-- ImageSets
|-- JPEGImages
|-- SegmentationClass
|-- SegmentationObject
- Modify the configuration file under the
experiments
folder according to your needs. - Change the
CONFIG
parameter insetup.py
, and then runsetup.py
to start training or detect multiple pictures at once.
运行结果(Results)
TODO list
- 训练和测试代码
- Yolo_v3
- Yolo_v4
- Yolo_v5
- Yolo_X
- CenterNet
- 发布在COCO数据集上训练好的模型
- Yolo_v3
- Yolo_v4
- Yolo_v5
- Yolo_X
- CenterNet
References
- https://github.com/hunglc007/tensorflow-yolov4-tflite
- https://github.com/amdegroot/ssd.pytorch code for processing COCO dataset
- https://github.com/ultralytics/yolov5
- https://github.com/calmisential/YOLOv4_PyTorch
- https://github.com/calmisential/CenterNet_TensorFlow2
- https://github.com/xingyizhou/CenterNet