Some Problem

[seulqxq]

Hello, I made a mistake when I used the demo example data set you provided for training, could you give me some advice? Thank you very much!

[wrld]

Sorry for the delayed reply. It seems like the shape of input depth to the depth loss is not correct, you can check the shape of the two inputs to make sure they have the shape like [H, W].

[NNsauce]

@seulqxq , hi you solved it? I also met this error
it comes from this two line:

obviously, mono_dep.shape is (1, 480, 640) and render_pkg["render_dep"].shape is (480, 640), they just miss a dimension.
I am not clear why the author has to do like that.
@wrld hi, about the two loss function:pearson_depth_loss and local_pearson_loss,
I modify them like below:
Am I right?

[NNsauce]

[wrld]

@NNsauce Your revision is right. This bug has been fixed, thank you for your comments.