Hey, thanks for this great project. I was trying to train Streetsurf on Segment-13476374534576730229_240_000_260_000, config = withmask_nolidar.240219.yaml, generated Depth Maps using Mini-omnidata(https://github.com/theNded/mini-omnidata) and had to make some changes to mono.py and train.py(line 180) = self.mono_depth_loss = MonoSDFDepthLoss(**loss_cfg['mono_depth'], debug_val_every=self.i_val). Changed the loss_type from pearson to monosdf in mono.py line 269. Right before the training started, I ran into the above error. I did not make any changes to the height or width of the images.
Hey, thanks for this great project. I was trying to train Streetsurf on Segment-13476374534576730229_240_000_260_000, config = withmask_nolidar.240219.yaml, generated Depth Maps using Mini-omnidata(https://github.com/theNded/mini-omnidata) and had to make some changes to
mono.py
andtrain.py
(line 180) = self.mono_depth_loss = MonoSDFDepthLoss(**loss_cfg['mono_depth'], debug_val_every=self.i_val). Changed the loss_type from pearson to monosdf inmono.py
line 269. Right before the training started, I ran into the above error. I did not make any changes to the height or width of the images.