Hi all,
I am trying to read files from a directory and use them later for a segmentation task.
I wrote the get_msk
function as suggested here (Practical-Deep-Learning-for-Coders)
The above function contains the line fn = path/f'{fn.stem}_seg.png'
, where I would select the masks, characterized by the suffix _seg.png'
Example:
image name = ./MyPath/4730043/43_tile10x11.jpg
mask name = ./MyPath/4730043/43_tile10x11_seg.png
When I run the dataloader dls = dblk.dataloaders('./MyDataset', bs=8)
I get an error because it can not find (at example) the file 43_tile10x11_seg_seg.png
where the suffix _seg is repeated.
I suspect that this kind of error is due to the aforementioned fn
path, but I am not sure about it.
How could I fix this issue?
Note: this the lambda function get_y = lambda o: get_msk(o, p2c)
applied by the dataloader.