项目作者: xiaochus

项目描述 :
Keras implementation of yolo v3 object detection.
高级语言: Python
项目地址: git://github.com/xiaochus/YOLOv3.git
创建时间: 2018-04-09T13:28:13Z
项目社区:https://github.com/xiaochus/YOLOv3

开源协议:MIT License

下载


YOLOv3

Keras(TF backend) implementation of yolo v3 objects detection.

According to the paper YOLOv3: An Incremental Improvement.

Requirement

  • OpenCV 3.4
  • Python 3.6
  • Tensorflow-gpu 1.5.0
  • Keras 2.1.3

Quick start

  • Download official yolov3.weights and put it on top floder of project.

  • Run the follow command to convert darknet weight file to keras h5 file. The yad2k.py was modified from allanzelener/YAD2K.

    1. python yad2k.py cfg\yolo.cfg yolov3.weights data\yolo.h5
  • run follow command to show the demo. The result can be found in images\res\ floder.

    1. python demo.py

Demo result

It can be seen that yolo v3 has a better classification ability than yolo v2.

TODO

  • Train the model.

Reference

  1. @article{YOLOv3,
  2. title={YOLOv3: An Incremental Improvement},
  3. author={J Redmon, A Farhadi },
  4. year={2018}

See LICENSE for details.