Closed Soulthym closed 5 years ago
Hi @Soulthym,
Yes, you can. You can check for the format your images need to have (in you case you will set the bounding box as the entire image): https://github.com/1adrianb/face-alignment/blob/master/face_alignment/detection/folder/folder_detector.py
Then, instead of passing the 'sfd' option for detection pass the above one. Alternatively the face alignment option support as an extra argument the bounding boxes directly(see: get_landmarks_from_image). Do note however that depending on the way you cropped the images this may result in an accuracy drop.
It works perfectly thanks! I ended up using your detector because it is way more accurate anyway, I hope this issue will help someone having the same question in the future :) Good job!
Hi, I am currently using another facial detector (based on YOLO) and would really need to get 3D landmarks on top of it, is it possible to use this module without the detector part? That would be greatly appreciated since I already have cropped faces and am running on a CPU