Invalid split name: train_aug

[katoabc123]

When I try to train the VOC2012 dataset. It show the error : ValueError: Invalid split name: train_aug

It happen in the voc.py file line 43:
raise ValueError("Invalid split name: {}".format(self.split))

I tried to set the root to the "/data/datasets/voc12/VOCdevkit" but it still cannot fix the problem.

What should I do?

[Kazuto Nakashima]

What is a value of your COFNIG.DATASET.NAME set in a config file? If you want to train on the train_aug set, please use vocaug not just voc.

deeplab-pytorch/main.py

Lines 118 to 128 in 79cb390

 dataset = get_dataset(CONFIG.DATASET.NAME)( 
     root=CONFIG.DATASET.ROOT, 
     split=CONFIG.DATASET.SPLIT.TRAIN, 
     ignore_label=CONFIG.DATASET.IGNORE_LABEL, 
     mean_bgr=(CONFIG.IMAGE.MEAN.B, CONFIG.IMAGE.MEAN.G, CONFIG.IMAGE.MEAN.R), 
     augment=True, 
     base_size=CONFIG.IMAGE.SIZE.BASE, 
     crop_size=CONFIG.IMAGE.SIZE.TRAIN, 
     scales=CONFIG.DATASET.SCALES, 
     flip=True, 
 ) 

[katoabc123]

Thank you very much. It works.