scnn_pytorch icon indicating copy to clipboard operation
scnn_pytorch copied to clipboard

Published 20 hours ago • verified publisher icon jcdubron

train

Open hellosher opened this issue 6 years ago • 4 comments

Can you provide the format of the data used for training?Or could you please prefect the readme file?

hellosher avatar Jul 22 '19 01:07 hellosher

Download CULane dataset and modify train_data_dir and eval_data_dir according to the root directory of CULane dataset.

jcdubron avatar Jul 22 '19 02:07 jcdubron

after i downloaded the dataset,there are some .txt files,including train.txt,train_gt.txt,val.txt,val_gt.txt,which files are needed in your code?sorry to bother you ,I don't quite understand your code

hellosher avatar Jul 22 '19 02:07 hellosher

These files are lists of images and segmentation ground truth, which are used to generate dataloaders.

jcdubron avatar Jul 22 '19 05:07 jcdubron

does the train.txt need to change to include the ground truth label,just like this "/driver_23_30frame/05171102_0766.MP4/00020.jpg /laneseg_label_w16/driver_23_30frame/05171102_0766.MP4/00020.png 1 1 1 0",especially the "1 1 1 0"?Or we still need to include those raw files?

hellosher avatar Jul 22 '19 06:07 hellosher