Hi, thank you for the great work and open-sourcing the code.
While running run_pre_process.py on a batch of high-quality fundus images, I noticed that no masks are generated, and the program silently skips many images. I think that the root cause is in the get_mask() function:
The call to cv2.HoughCircles() in _get_center_radius_by_hough() often fails (returns None) even on high-quality images, especially when the thresholding in get_mask_BZ() produces weak or broken masks.
Since the failure is silently caught by a try/except block, no error is shown, and no output is saved for the skipped images.
Hi, thank you for the great work and open-sourcing the code.
While running
run_pre_process.pyon a batch of high-quality fundus images, I noticed that no masks are generated, and the program silently skips many images. I think that the root cause is in theget_mask()function:The call to
cv2.HoughCircles()in_get_center_radius_by_hough()often fails (returns None) even on high-quality images, especially when the thresholding inget_mask_BZ()produces weak or broken masks.Since the failure is silently caught by a
try/exceptblock, no error is shown, and no output is saved for the skipped images.