About the image crop

[shaohua-pan]

Hi ！Thanks for the good job！ I found that in the imutils.py line 77: new_img[new_y[0]:new_y[1], new_x[0]:new_x[1]] = img[old_y[0]:old_y[1], old_x[0]:old_x[1]]. If I use this, I will get the crop picture for the first one. And I change it to: new_img = img[old_y[0]:old_y[1], old_x[0]:old_x[1]]. The result will be the second one. I am just wondering which is the correct input? Not the second one? Thanks a lot.

[shaohua-pan]

I just use cv2.imshow('frame1', cv2.cvtColor(new_img, cv2.COLOR_BGR2RGB)) cv2.waitKey(1) to show the image.

[zhihaolee]

The first one should be the correct input. The visualization problem could be about the data type.

[shaohua-pan]

Thanks. What is the difference between these？I I I understand the network input is a crop image with normalization. What is the first one meaning or what is the problem about data type（I just use cv2.imshow）

[zhihaolee]

The visualization could be different for different input data types, like uint8 and float32.