roytseng-tw / adversarial-pose-pytorch

A PyTorch implementation of adversarial pose estimation for multi-person

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Adversarial Pose Estimation

This repository implements pose estimation methods in PyTorch.

Getting Started

Data

The file lsp_mpii.h5 contains the annotations of MPII, LSP training data and LSP test data.
Place LSP, MPII images in data/LSP/images and data/mpii/images.
Place coco annotations in data/coco/annotations and images in data/coco/images, as suggested by cocoapi. The file valid_id contains the image_ids used for validation.

Compile the extension

Compile the C implementation of the associative embedding loss. Code credit umich-vl/pose-ae-train.

cd src/extensions/AE
python build.py  # be sure to have visible cuda device

Folder Structure

  • data: put the training / testing data here
  • src:
    • models: model definition
    • datasets: dataset definition
    • extensions:
    • utils

All the other folders represents different tasks. Each contains a training script train.py and definition of command-line options opts.py.

Known Issues

  • advpose-ae: Only supports single gpu. Multi-gpu training get stucked randomly. The problem seems to be caused by the AE_loss extension.

TODOs

  • visualization
  • example of usage

About

A PyTorch implementation of adversarial pose estimation for multi-person


Languages

Language:Python 91.5%Language:C 8.0%Language:Cuda 0.3%Language:C++ 0.2%