Skip to content

Final project for CS 7643 Deep Learning focusing on reproducing the results detailed in "Self-Supervised Monocular Scene Decomposition and Depth Estimation" - S. Safadoust, F. Guney (2021)

Notifications You must be signed in to change notification settings

ShuvoNewaz/MonoDepthSegReprod

 
 

Repository files navigation

MonoDepthSegReprod

Final project for CS 7643 Deep Learning focusing on reproducing the results detailed in "Self-Supervised Monocular Scene Decomposition and Depth Estimation" - S. Safadoust, F. Guney (2021)

Contributors

Luis Pimentel

Nicholas Cich

Shuvo Newaz

Eric Gu

Dependencies (Current)

  • torchvision-0.11.1
  • pytorch-1.10.0
  • numpy-1.21.2

Training and Validation

Command for Training:

  • python train.py --png --data_path ../kitti_data --load_weights_folder /$path_to_weights$

Command for Evaluation:

  • python depth_masking_validation.py --png --data_path /$path_to_kitti_data$ --load_weights_folder /$path_to_weights$ --num_val_batches=2

Command for Viewing Logs:

  • tensorboard --logdir=<dir_of_log>

References

This codebase reproduces the MonoDepthSeg architecture described in the following work:

Self-Supervised Monocular Scene Decomposition and Depth Estimation

Sadra Safadoust, Fatma Güney

3DV 2021 (arXiv pdf)

No public code implementations for this work currently exist.


This codebase also uses model architectures and utility functions from the following work per their non-commercial use license:

Digging into Self-Supervised Monocular Depth Prediction

Clément Godard, Oisin Mac Aodha, Michael Firman and Gabriel J. Brostow

ICCV 2019 (arXiv pdf)

nianticlabs/monodepth2 (Github)

Copyright © Niantic, Inc. 2018. Patent Pending.

About

Final project for CS 7643 Deep Learning focusing on reproducing the results detailed in "Self-Supervised Monocular Scene Decomposition and Depth Estimation" - S. Safadoust, F. Guney (2021)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%