Why is the splitting of train dataset hardcoded?

[GowthamGottimukkala]

While dividing the train set into two classes, why is it hardcoded that first 63 values are one class and the remaining are one? Aren't we supposed to split it using the labels file provided through option.py

RTFM/main.py

Lines 19 to 24 in 950243a

	train_nloader = DataLoader(Dataset(args, test_mode=False, is_normal=True),
	batch_size=args.batch_size, shuffle=True,
	num_workers=0, pin_memory=False, drop_last=True)
	train_aloader = DataLoader(Dataset(args, test_mode=False, is_normal=False),
	batch_size=args.batch_size, shuffle=True,
	num_workers=0, pin_memory=False, drop_last=True)

RTFM/dataset.py

Lines 23 to 34 in 950243a

	def _parse_list(self):
	self.list = list(open(self.rgb_list_file))
	if self.test_mode is False:
	if self.is_normal:
	self.list = self.list[63:]
	print('normal list')
	print(self.list)
	else:
	self.list = self.list[:63]
	print('abnormal list')
	print(self.list)

[tianyu0207]

While dividing the train set into two classes, why is it hardcoded that first 63 values are one class and the remaining are one? Aren't we supposed to split it using the labels file provided through option.py

RTFM/main.py

Lines 19 to 24 in 950243a

	train_nloader = DataLoader(Dataset(args, test_mode=False, is_normal=True),
	batch_size=args.batch_size, shuffle=True,
	num_workers=0, pin_memory=False, drop_last=True)
	train_aloader = DataLoader(Dataset(args, test_mode=False, is_normal=False),
	batch_size=args.batch_size, shuffle=True,
	num_workers=0, pin_memory=False, drop_last=True)

RTFM/dataset.py

Lines 23 to 34 in 950243a

	def _parse_list(self):
	self.list = list(open(self.rgb_list_file))
	if self.test_mode is False:
	if self.is_normal:
	self.list = self.list[63:]
	print('normal list')
	print(self.list)
	else:
	self.list = self.list[:63]
	print('abnormal list')
	print(self.list)

Yes. I got lazy and hard-coded the weak label for the training set. For the array [:63], all the videos are abnormal and the remaining are normal. This is just a function to provide abnormal/normal list for the dataloader so that I can evenly sample 32 videos for each batch from both classes.

[GowthamGottimukkala]

1. Ok thanks, I'll provide another argument for an input file that has weak labels for the train set along with the already existing test label file argument.
2. By the way, does that mean the shanghai tech i3d features (list/shanghai-i3d-train-10crop.list) you used have first 64 lines as abnormal ones and the rest are normal ones since you hardcoded it?

[tianyu0207]

1. Ok thanks, I'll provide another argument for an input file that has weak labels for the train set along with the already existing test label file argument.
2. By the way, does that mean the shanghai tech i3d features (list/shanghai-i3d-train-10crop.list) you used have first 64 lines as abnormal ones and the rest are normal ones since you hardcoded it?

2. Yes. This is the same setup as the GCN-anomaly paper.