Hybrid Dynamic-static Context-aware Attention Network for Action Assessment in Long Videos

This is the code for ACMMM 2020 paper 'Hybrid Dynamic-static Context-aware Attention Network for Action Assessment in Long Videos'.

The code should be directly runnable with PyTorch v1.0.1.

Rhythmic Gymnastics Dataset

The Rhythmic Gymnastics Dataset can be downloaded from here.

The extracted features of Rhythmic Gymnastics Dataset can be downloaded from here.

Name		Name	Last commit message	Last commit date
Latest commit History 8 Commits
data		data
models		models
pretrained_models		pretrained_models
.gitignore		.gitignore
README.md		README.md
conf.py		conf.py
dataset.py		dataset.py
extract_dynamic_feat.py		extract_dynamic_feat.py
extract_static_feat.py		extract_static_feat.py
main.py		main.py
trainer.py		trainer.py
utils.py		utils.py