pseudo_depth in the data_modules.py

[chunqizhang]

Dear Dr.Bian,
I train on my dataset with --val_mode photo, but I find the with_pseudo_depth=False in the data_modules. If it is False, will the pseudo_depth generated by the Leres effect?
Thank you very much for your patient reply!

elif self.hparams.hparams.val_mode == 'photo':
            self.val_dataset = TrainFolder(
                self.hparams.hparams.dataset_dir,
                train=False,
                transform=self.valid_transform,
                sequence_length=self.hparams.hparams.sequence_length,
                skip_frames=self.hparams.hparams.skip_frames,
                use_frame_index=self.hparams.hparams.use_frame_index,
                with_pseudo_depth=False
            )

[JiawangBian]

Pseudo-depth is only used for training. Here, for validation, I only use photometric loss, so the pseudo-depth is not used.

[chunqizhang]

Pseudo-depth is only used for training. Here, for validation, I only use photometric loss, so the pseudo-depth is not used.

Thank you for your reply!