search

noahzn / Lite-Mono

[CVPR2023] Lite-Mono: A Lightweight CNN and Transformer Architecture for Self-Supervised Monocular Depth Estimation
MIT License
540 stars 61 forks source link

self.models_pose["pose_encoder"] or self.models["pose_encoder"]? #42

Closed howardchina closed 1 year ago

howardchina commented 1 year ago

A small typo I guess.

https://github.com/noahzn/Lite-Mono/blob/286c4ab750ad491f0930791b29e401e4882f20d9/trainer.py#L343

noahzn commented 1 year ago

Hi, Yes, you are correct! However, pose_model_input is set to pairs, so self.num_pose_frames = 2 and this line of code will never be executed. If you set pose_model_input=all you then need to fix this typo. Thank you!

howardchina commented 1 year ago

Yes, pairs mode will never call this line.