using mmselfsup models for classification and activity recognition

[alaa-shubbak]

Checklist

1. I have searched related issues but cannot get the expected help. yes
2. I have read the FAQ documentation but cannot get the expected help. yes

I am interested to know if i can use the structure of models in this library to design a self-supervised classifier and activity recognition model for a time-series dataset of sequence frames(images).

what are your suggestions to improve the model and make it able to do contrastive self-supervised learning on sequences of images, instead of a single image? shall i use the self-supervised stage for images, then work on the sequence for the downstream classification task only?