Call the model's compute loss function Currently only supports supervised methods (ENet, PENet, MSGCHN, NLSPN) Unsupervised methods have various more complex losses that is best trained through their repository Arg(s): output_depth : torch.Tenso
(self, output_depth, ground_truth_depth)
source not stored for this graph (policy: none)
nothing calls this directly
no outgoing calls
no test coverage detected