The pre-process function prepare_input pads the image when it contains chunks of smaller size than the input's chunk size. These usually are located at the tailing edges of the image.
Therefore the output labels image has different spatial shape than the input image. This can be solved by removing the corresponding padded pixels from the labels output.
fercer
commented
6 months ago
The pre-process function
prepare_inputpads the image when it contains chunks of smaller size than the input's chunk size. These usually are located at the tailing edges of the image.Therefore the output labels image has different spatial shape than the input image. This can be solved by removing the corresponding padded pixels from the labels output.