mayuelala/aku

[ACM MM 2022 Oral] AKU: This repo is the official implementation of "Visual Knowledge Graph for Human Action Reasoning in Videos"

🚀 [ ACM MM 2022 Oral ] AKU: Action Knowledge Understanding framework

This is the official repository of our ACM Multimedia 2022 Work. Kinetic-TPS dataset and pytorch training/validation code for Visual Knowledge Graph for Human Action Reasoning in Videos.

🏚️ Introduction

🏝️ Visualizations

🏰 Acknowledgement

This work is partially supported by National Natural Science Foun- dation of China (61876176), the Joint Lab of CAS-HK, the Shanghai Committee of Science and Technology (Grant No. 21DZ1100100).

aku's Issues

How does it work for new action videos?

I thing that your model goes like this,
you have kinetic tps annotated data which used to build a knowledge graph,
now what about non annotated data will you first , use cnn or something for creating those part, verb tags and then procced to do the knowledge graph thing or any other thing. please consider this question and do answer.

哭了

Can there be more sincerity between people?

Code

I am trying to learn how your model works in details, for an in depth understanding i would like to ponder upon those codes.

Waiting for you to open more resources

This is a very interesting project.