princeton-vl/RAFT-Stereo

Parameter missing "args.context_norm"

Opened this issue · 1 comments

Thanks for your nice work, when I use "real_time_model" for inference, class RAFTStereo requires norm_fn=args.context_norm, but the error args seems to not exist in this parameter.

hi,
Do you have fixed this problem now? I meet the same problem with you.
THX

Traceback (most recent call last):
File "test_prototype.py", line 140, in
disparity = disparity_from_stereo_raft2d(img_rect255., img_rect_2255., 'real_time_model', device)
File "F:\Test\Test-neualpassthrough\NeuralPassthrough\code\stereo_depth.py", line 61, in disparity_from_stereo_raft2d
model = torch.nn.DataParallel(RAFTStereo(args), device_ids=[0])
File "F:\Test\Test-neualpassthrough\NeuralPassthrough\code\third_party\RAFT_Stereo\core\raft_stereo.py", line 29, in init
self.cnet = MultiBasicEncoder(output_dim=[args.hidden_dims, context_dims], norm_fn=args.context_norm, downsample=args.n_downsample)
AttributeError: 'Namespace' object has no attribute 'context_norm'