Open aberenguel opened 3 weeks ago
I think some changes could address this issue:
1) Convert the face_encodings
from byte[] list to NDArray list in IPEDReader
2) Set face_count = 0
when no faces were found.
3) Check in FaceRecognitionTask.py if face_count
is set. If so, skip it.
I processed a case with
enableFaceRecognition = true
. When I generated a report, it seems the FaceRecognitionTask process the images again. Since, the indexed items already have the face recognition extra properties, it could be reused.