some questions and a note

[hamedsteiner]

Hi, thanks for the great project. I wanted more clarification on the state (score) output for each face in an image, as there wasn't much documentation about it.
Also for multiface inference for images, the model seems to have an issue which can be fixed by changing onnx GraphOptimizationLevel