li-jicheng / mmasd-a-multimodal-dataset-for-autism-intervention-analysis Goto Github PK

Hi, I was looking into your dataset (2D openpose). The JSON file provides 3 persons. Is there a way to identify the person with autism or is it random in the JSON structure?
In case it's random, are the positions of the trainer, model, and autistic child always the same as shown in figure 1 of the paper?
Another question is how to interpret the folder names in each activity class and relate it to the demographics (ID) provided in the csv file.

Thanks.