Extracting skeleton data using openpose

[lakpa-tamang9]

In the old stgcn implementation, there is a offline_demo.py script to run demo on the trained model. However, i want following:

• Create custom dataset using openpose
• Train the data using old stgcn implementation
• Run demo on videos using offline_demo.py

In the old stgcn, it they have provided the google drive link to the preprocessed dataset. However, it is not mentioned how do they prepare data using openpose. I want to know this @yysijie, please look into this issue. This is important because, if i want my custom dataset to be trained and inferenced, then i should know how the data is being prepared at the first place.