search

AayushKrChaudhary / RITnet

This is a winning model of OpenEDS Semantic Segmentation Challenge
MIT License
67 stars 27 forks source link

RITnet retraining problem #4

Closed Leesisisi closed 3 years ago

Leesisisi commented 3 years ago

Hello, this is Lee, I‘m very interested in the network you published. I now have the data set I collected and want to retrain on your network, but some problems occurred. I changed the instruction according to the name and path of my data set, but still reported an error: Traceback (most recent call last): File "train.py", line 103, in transform = transform, **kwargs) File "/home/mprl-et/leesisisi/RITnet-master/dataset.py", line 131, in init for file in os.listdir(osp.join(self.filepath,'images')):
FileNotFoundError: [Errno 2] No such file or directory: 'RITnet_Self_Dataset/train/images'

I'm sure the filepath is right, and because of our picture format is jpg, I changed the dataset.py line132 and 133: for file in os.listdir(osp.join(self.filepath,'images')):
if file.endswith(".jpg"): listall.append(file.strip(".jpg"))

Will this cause any problems? Are there specific requirements for the data format of the training set? If so please let me know.

Thank you.

AayushKrChaudhary commented 3 years ago

That should fix the issue. Can you print and check if listall is empty or contains all the images you want?

AayushKrChaudhary commented 3 years ago

Is more of the folder arrangement issue rather than a code base issue. So closing this issue.

Folder should be arranged in Folder_name/train(test)/images/ If you are using '.jpg' change lines 132,133,148 of file dataset.py [Updated on Sept 19, 2019]