Vandmoon

Just a reminder: The current implementation of build_inceptionresnetv1 in dev branch seems problematic, althrough not finished yet, since the input of the FaceVerifier(extractor="facenet", classes=512) is assumed to be an image...

The "max" should be "min" in the line: https://github.com/poodarchu/learn_aug_for_object_detection.numpy/blob/535645b9b992fead98f7aafdf0d86d674edd0025/autoaugment_utils.py#L447