current yolo logic is not working for 8.0.122

[serengil]

Laird Foret raised this issue. He is using yolov8 with exact version '8.0.122'

Current code block:

        if align:
            # Extract landmarks
            left_eye, right_eye, _, _, _ = result.keypoints.tolist()
            # Check the landmarks confidence before alignment
            if (left_eye[2] > LANDMARKS_CONFIDENCE_THRESHOLD and
                    right_eye[2] > LANDMARKS_CONFIDENCE_THRESHOLD):
                detected_face = FaceDetector.alignment_procedure(
                    detected_face, left_eye[:2], right_eye[:2]
                )

He mentioned this should be used not to get any exception.

        if align:
            # Extract landmarks
            keypoints_xy = result.keypoints.xy  # Extract the keypoints coordinates
            keypoints_conf = result.keypoints.conf  # Extract the confidence values for keypoints
            left_eye = (keypoints_xy[0][0], keypoints_conf[0][0])  # Tuple of x,y and confidence for left eye
            right_eye = (keypoints_xy[0][1], keypoints_conf[0][1])  # Tuple of x,y and confidence for right eye

            # Check the landmarks confidence before alignment
            if (left_eye[1] > LANDMARKS_CONFIDENCE_THRESHOLD and
                    right_eye[1] > LANDMARKS_CONFIDENCE_THRESHOLD):
                detected_face = FaceDetector.alignment_procedure(
                    detected_face, left_eye[0].cpu(), right_eye[0].cpu()
                )

@Vincent-Stragier you may have a comment. Besides, why we used specific github version instead of pip install ultralytics

[Vincent-Stragier]

Hello,

I suppose there are some differences between the data structures used in the different version. I personally think it would be better to migrate to the original Ultralytics source. However, I'd simplified the changes a bit:

        if align:
            # Tuple of x, y and confidence for left eye
            left_eye = result.keypoints.xy[0][0], result.keypoints.conf[0][0]
            # Tuple of x, y and confidence for right eye
            right_eye = result.keypoints.xy[0][1], result.keypoints.conf[0][1]

            # Check the landmarks confidence before alignment
            if (left_eye[1] > LANDMARKS_CONFIDENCE_THRESHOLD and
                    right_eye[1] > LANDMARKS_CONFIDENCE_THRESHOLD):
                detected_face = FaceDetector.alignment_procedure(
                    detected_face, left_eye[0].cpu(), right_eye[0].cpu()
                )

It is slightly faster.

[serengil]

would you please create a PR for this?

[Vincent-Stragier]

#790 should do it.