Skip to content

Depth Reconstruction Research: interpolation and super-resolution of depth maps using deep convolutional neural networks.

Notifications You must be signed in to change notification settings

slavija-b/depth_reconstruction

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

16 Commits
 
 
 
 
 
 
 
 

Repository files navigation

depth_reconstruction

Depth Reconstruction Research

  1. The fast interpolation method is composed of the interpolating and super-resolving networks described in the paper "Semi-Dense Depth Interpolation using Deep Convolutional Neural Networks" by Ilya Makarov, Vladimir Aliev and Olga Gerasimova. The original code has been written by Vladimir Aliev and used here for research purposes.

  2. The code for the super-resolving network is written based on the TensorFlow implementation of "Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network" (https://github.com/tensorlayer/srgan).

  3. The code for unfinished research on interpolation guided by edges is also added.

[NB] The code in this repo DOES NOT describe the finished product but mostly represents the experiments and research materials. Please refer to the papers for the exact parameters!

[NB] VGG weights are used in all these networks.

Research works based on these resources:

  1. Fast Semi-dense Depth Map Estimation https://dl.acm.org/citation.cfm?doid=3210499.3210529
  2. Super-resolution of interpolated downsampled semi-dense depth map https://dl.acm.org/citation.cfm?id=3208821
  3. Sparse Depth Map Interpolation using Deep Convolutional Neural Networks https://ieeexplore.ieee.org/document/8441443/
  4. Fast depth map super-resolution using deep neural network https://www.ismar2018.org/papers/ismar2018_pcs_poster_1185.html

About

Depth Reconstruction Research: interpolation and super-resolution of depth maps using deep convolutional neural networks.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 97.0%
  • Python 3.0%