Yuliang (Zack) Zou

@isheetajha Due to some system update, I cannot get access to the old code base. I will share the results soon.

@roboticsbala You can use val.txt to test the depth estimation performance for model selection. Simply replace test.txt with val.txt

@isheetajha I actually used the simple L2 photometric loss for the pre-training, since I found it easier to train the network.

@isheetajha Did you use val set to pick the best model? Or did you monitor the training progress using tensorboard visualization? The training loss will not decrease obviously in the...

@isheetajha The pixel loss curve looks good, but the smoothness loss curve clearly indicates that the training fails (since it decreases to zero quickly, if you observe the predicted depth...

Yes, definitely. I just got back to campus and would take a look during the weekend.

Hi @xy-guo , is the performance much worse than the reported number? If you don't mind, can you share your implementation of this part? Thanks.

Also waiting for this model!

I think you have to use GPU for it since some operations are GPU-specific.

Hi @PK15946 , thanks for the information. Do you want to make a pull request?