Skip to content

zhengfeiwang/multi-modal-action-classification

Repository files navigation

multi-modal-action-classification

Action classification based on deep learning with multi modal datasets.

Table of Contents

Prerequisites

Dependencies is required as follow:

  • PyTorch 1.0
  • tensorboardX 1.6 and tensorboard 1.12

Skeleton

Implementation of HCN (Co-occurrence Feature Learning from Skeleton Data for Action Recognition and Detection with Hierarchical Aggregation. Hikvision Research Institute. IJCAI 2018. [arXiv])

preprocess

# change the access permissions
chmod 777 preprocess_script.sh
# preprocess dataset using arguments
# 1. PKU-MMD version, 2. sequence length
./preprocess_script.sh 1 256

Training

skeleton

# if has GPUs, add arguments --gpu
python main.py --sequence-length <SEQUENCE_LENGTH>

About

Action classification based on multi-modal dataset

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages