search

610265158 / Peppa_Pig_Face_Landmark

A simple face detect and alignment method, which is easy and stable.
Apache License 2.0
530 stars 116 forks source link

Do you need assistance to complete your work? #48

Open ofirkris opened 1 year ago

ofirkris commented 1 year ago

Hello, great effort on this implementation.

Suggestion - updating face detection to use YoloNas for facial detection. Is the Main Teacher model outperforming Spiga? Would you need GPU resources / other assistance to complete your work? How far are you from achieving SoTa results?

610265158 commented 1 year ago

Hi, thanks for the suggestion. For face detection, I think YoloNas will be a good model, but i have no pan to train it for now.

The teacher model achieves a little better than Spiga, however i am not familiar with Spiga.

Gpu resources is not a key problem here, because dataset is small. However, with more compute resources, the experiments will run faster, and also, bigger model should achieves better result.

I am still running the model with some params tunning. And maybe testing on other datasets with the same model.