BoundingBoxAnnotatorCrop ignores the selected bounding boxes
The BoundingBoxAnnotatorCrop
is nice in the sense that it can even work when no face was detected by simply using the bounding box. While the "topleft"
and "bottomright"
are required to be specified, they are however ignored in the constructor:
https://gitlab.idiap.ch/bob/bob.bio.face/-/blob/ead8c069bafb4024dc15c5df7fdc878aec8bd5f0/bob/bio/face/preprocessor/FaceCrop.py#L542
Instead, the face is cropped and simply scaled to the required size: https://gitlab.idiap.ch/bob/bob.bio.face/-/blob/ead8c069bafb4024dc15c5df7fdc878aec8bd5f0/bob/bio/face/preprocessor/FaceCrop.py#L616
Finally, the fixed_positions
are not respected by the annotator, so it is impossible to use it with a dataset that does not provide topleft
and bottomright
annotations.