hujc7 / yolo-v3

You Only Look Once by Pytorch

Introduction

YOLO-v3 implemention from "YOLOv3: An Incremental Improvement".

Tutorial

Get tutorial series in HomePage or ZhihuPage if know Chinese

Requirement

• python 3.6
• pytorch 0.4.0
• numpy 1.13.1
• PIL

Usage

Options

usage: detect.py [-h] [--images IMAGES] [--result RESULT]
                 [--batch_size BATCH_SIZE] [--img_size IMG_SIZE]
                 [--confidence CONFIDENCE] [--nms_thresh NMS_THRESH]
                 [--weights WEIGHTS] [--no_cuda]

YOLO-v3 Detect

optional arguments:
  -h, --help            show this help message and exit
  --images IMAGES
  --result RESULT
  --batch_size BATCH_SIZE
  --img_size IMG_SIZE
  --confidence CONFIDENCE
  --nms_thresh NMS_THRESH
  --weights WEIGHTS
  --no_cuda

Detect

Download weight from GoogleDrive or BaiduYunPan to $(PROJECT_HOME)/

python3 detect.py

Citation

If you find this code useful for your research, please cite:

@misc{TaoYOLOv3,
  author = {Ne7ermore Tao},
  title = {yolo-v3},
  publisher = {GitHub},
  year = {2018},
  howpublished = {\url{https://github.com/ne7ermore/yolo-v3}}
}

Contact

Feel free to contact me if there is any question (Tao [email protected]).