Currently, the whole image is dropped if the cell is too close at the border of the image (such that the requested ROI would go beyond the original image).
This can be a problem for small datasets where each event counts.
The function "image_crop_pad_cv2" in
https://github.com/maikherbig/AIDeveloper/blob/master/Tutorial%20Deploy%20to%20OpenCV%20dnn/aid_cv2_dnn.py
already solved that issue. Hence, just use that in aid_img.py!
Currently, the whole image is dropped if the cell is too close at the border of the image (such that the requested ROI would go beyond the original image). This can be a problem for small datasets where each event counts. The function "image_crop_pad_cv2" in https://github.com/maikherbig/AIDeveloper/blob/master/Tutorial%20Deploy%20to%20OpenCV%20dnn/aid_cv2_dnn.py already solved that issue. Hence, just use that in aid_img.py!
Thanks to Martin Kräter (https://twitter.com/KraterMartin) for reporting!