CGCN: Context Graph Convolutional Network for Few-Shot Temporal Action Localization

Introduction

This is a PyTorch implementation for our paper "CGCN: Context Graph Convolutional Network for Few-Shot Temporal Action Localization".

Dependencies

• Python == 3.7
• Pytorch==1.1.0 or 1.3.0
• CUDA==10.0.130
• CUDNN==7.5.1_0
• GCC >= 4.9
• pip install git+https://github.com/luizgh/visdom_logger.git

Installation

1. Create conda environment

   conda env create -f env.yml
   source activate cgcn

2. Install Align1D2.2.0

   cd cgcn_lib
   python setup.py install

3. Test Align1D2.2.0

   python align.py

4. Post-processing : Download the CUHK classifier from this link and place it in "data" folder

License

This project is licensed under the Apache-2.0 License.