search

machanic / AU_R-CNN

The official implementation code of paper: "AU R-CNN：Encoding Expert Prior Knowledge into R-CNN for Action Unit Detection".
https://arxiv.org/abs/1812.05788
87 stars 24 forks source link

Is there any way to assess AUs of single video or image? #17

Open iPsych opened 4 years ago

iPsych commented 4 years ago

Hi,

Is there any way to assess AUs of single video or image, not a test-dataset? Any recommended way to pre-process image/video (cropping, face bounding box generation), and the code to start with?

machanic commented 4 years ago
  1. Yes, we can assess the performance of a single image. Look at my evaluation code
  2. You can use Dlib to extract all the landmarks of a face. Then you can crop the face to exclude the background area using the landmark point around the face edge. My code have provide this step in https://github.com/machanic/AU_R-CNN/blob/master/img_toolkit/face_mask_cropper.py#L63