Estimation of monocular depth using CNN

Introduction

Our model is based on similar work Towards Good Practice for CNN-Based Monocular Depth Estimation. Our aim was to recreate given research on NYUv2 labeled dataset containing 1449 pictures.

Presentation

Link to pdf presentation

Important files

Changing hyperparameters:
hyperparameters.py

Training:
train.py

Test:
test.py

Test on sample image:
test_sample.py

Important notes!

• This project was part of Petnica's Summer Institute for Machine Learning 7, done by Mihailo Grbic and me.
• This project is cloned from Mihailo's repository