https://github.com/uber-research/DeepPruner

DeepPruner: Learning Efficient Stereo Matching via Differentiable PatchMatch (ICCV 2019)
https://github.com/uber-research/DeepPruner

iccv2019 patchmatch pytorch real-time stereo-matching stereo-vision

Last synced: 11 days ago
JSON representation

DeepPruner: Learning Efficient Stereo Matching via Differentiable PatchMatch (ICCV 2019)

• Host: GitHub
• URL: https://github.com/uber-research/DeepPruner
• Owner: uber-research
• License: other
• Created: 2019-10-15T23:48:59.000Z (about 5 years ago)
• Default Branch: master
• Last Pushed: 2020-09-05T08:17:08.000Z (about 4 years ago)
• Last Synced: 2024-08-01T20:36:55.836Z (3 months ago)
• Topics: iccv2019, patchmatch, pytorch, real-time, stereo-matching, stereo-vision
• Language: Python
• Homepage:
• Size: 19.3 MB
• Stars: 349
• Watchers: 17
• Forks: 41
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-computer-vision-papers - 2019ICCV

README

        
# DeepPruner: Learning Efficient Stereo Matching via Differentiable PatchMatch

This repository releases code for our paper [DeepPruner: Learning Efficient Stereo Matching via Differentiable PatchMatch](https://arxiv.org/abs/1909.05845).

##### Table of Contents  

[DeepPruner](#DeepPruner)  

[Differentiable Patch Match](#DifferentiablePatchMatch)  

[Requirements (Major Dependencies)](#Requirements)  

[Citation](#Citation)  

  

  ###  **DeepPruner** 

 + An efficient "Real Time Stereo Matching" algorithm, which takes as input 2 images and outputs a disparity (or depth) map.

	

	

	

	![](readme_images/DeepPruner.png)

	

	

	

 + Results/ Metrics:

							

    + [**KITTI**](http://www.cvlibs.net/datasets/kitti/eval_scene_flow.php?benchmark=stereo): **Results competitive to SOTA, while being real-time (8x faster than SOTA). SOTA among published real-time algorithms**.

	

    

    ![](readme_images/KITTI_test_set.png)

	  ![](readme_images/CRP.png)

    ![](readme_images/uncertainty_vis.png)

	

	

	  + [**ETH3D**](https://www.eth3d.net/low_res_two_view?mask=all&metric=bad-2-0): **SOTA among all ROB entries**. 

	  

    + **SceneFlow**: **2nd among all published algorithms, while being 8x faster than the 1st.**

	  

    


    

	  
 

         

    

	

	  + [**Robust Vision Challenge**](http://www.robustvision.net/index.php): **Overall ranking 1st**. 

	   

    


    

	  

	

	+ Runtime: **62ms** (for DeepPruner-fast), **180ms** (for DeepPruner-best)

	  

	+ Cuda Memory Requirements: **805MB** (for DeepPruner-best)


  

  ### **Differentiable Patch Match**

  + Fast algorithm for finding dense nearest neighbor correspondences between patches of images regions. 

    Differentiable version of the generalized Patch Match algorithm. ([Barnes et al.](https://gfx.cs.princeton.edu/pubs/Barnes_2010_TGP/index.php))

    

   


    

    


More details in the corresponding folder README.



## Requirements (Major Dependencies)

+ Pytorch (0.4.1+)

+ Python2.7

+ torchvision (0.2.0+)



## Citation

If you use our source code, or our paper, please consider citing the following:

> @inproceedings{Duggal2019ICCV,  

title = {DeepPruner: Learning Efficient Stereo Matching  via Differentiable PatchMatch},  

author = {Shivam Duggal and Shenlong Wang and Wei-Chiu Ma and Rui Hu and Raquel Urtasun},  

booktitle = {ICCV},  

year = {2019}

}

Correspondences to Shivam Duggal , Shenlong Wang , Wei-Chiu Ma