项目作者: Xiaoccer

项目描述 :
This is the python implementation of the datalayer in Convolutional Pose Machines
高级语言: Python
项目地址: git://github.com/Xiaoccer/CPM_PyLayer.git
创建时间: 2018-06-26T08:33:28Z
项目社区:https://github.com/Xiaoccer/CPM_PyLayer

开源协议:

下载


CPM_PyLayer

This is the python implementation of the datalayer in Convolutional Pose Machines release

Environments

  • Python2.7
  • Opencv2.4.13
  • Numpy
  • IPython

Deitals

data_transformer.py is the implenentaiton like the data_transformer.cpp in the Convolutional Pose Machines release code. But something is difference:

  • keep the method of data augmentation
  • reduce some params
  • remove the center map(if you want to detect mutli person, you can add it)

cpm_data.py shows how to use data_transformer.py in caffe.

util.py is the Visualization tools, produced by the Convolutional Pose Machines release

How to use

  • Replace the data_dir to your dataset
  • Refer the data format in example folder to generate your dataset

Cited Convolutional Pose Machine

  1. @inproceedings{wei2016cpm,
  2. author = {Shih-En Wei and Varun Ramakrishna and Takeo Kanade and Yaser Sheikh},
  3. booktitle = {CVPR},
  4. title = {Convolutional pose machines},
  5. year = {2016}
  6. }