kazumaohashimkb/stereonet-activestereonet

stereo matching StereoNet: Guided Hierarchical Refinement for Real-Time Edge-Aware Depth prediction model in pytorch. ECCV2018; ActiveStereoNet:End-to-End Self-Supervised Learning for Active Stereo Systems ECCV2018 Oral

StereoNet: Guided Hierarchical Refinement for Real-Time Edge-Aware Depth prediction model in pytorch. ECCV2018

ActiveStereoNet:End-to-End Self-Supervised Learning for Active Stereo Systems ECCV2018 Oral

If you want to communicate with me about the StereoNet, please concact me without hesitating. My email:

[email protected]

my model result

Now, my model's speed can achieve 25 FPS on 540*960 img with the best result of 1.87 EPE_all with 16X multi model， 1.95 EPE_all with 16X single model on sceneflow dataset by end-to-end training. the following are the side outputs and the prediction example

train example

test example

over 100FPS on titan xp gpu

• KITTI2015 submission: http://www.cvlibs.net/datasets/kitti/eval_scene_flow_detail.php?benchmark=stereo&result=94cf8e1c4f9d61df98c8e8cc8c726a34360208be ranking 179

real time application

point cloud view example

Citation

• refercence[1]

If you find our work useful in your research, please consider citing:

@inproceedings{khamis2018stereonet, title={Stereonet: Guided hierarchical refinement for real-time edge-aware depth prediction}, author={Khamis, Sameh and Fanello, Sean and Rhemann, Christoph and Kowdle, Adarsh and Valentin, Julien and Izadi, Shahram}, booktitle={Proceedings of the European Conference on Computer Vision (ECCV), Munich, Germany}, pages={8--14}, year={2018} }

Introduction

I implement the real-time stereo model according to the StereoNet model in pytorch. The speed can reach 30FPS with top performance. The speed can reach 60FPS with lower performance.

Method	EPE_all on sceneflow dataset	EPE_all on kitti2012 dataset	EPE_all on kitti2015 dataset
ours(8X single)	stage0:2.26 stage1:1.38
Reference[1]	stage1: 1.525

License

• Our code is released under MIT License (see LICENSE file for details).

Installaton

• python3.6
• pytorch0.4

Usage

• run main8Xmulti.py

Updates

• finetune the performance beating the original paper.

rethink

• Do not design massive deep networks with multiple stages to improve kitti by 1%(no meaning doing this)
• Use metrics that matter for visual navigation (hint: not L1 depth error)
• ...

pretrain model

• Sceneflow pretrain weight https://drive.google.com/open?id=1bSwewxrRfmFCxZDyAtyYyQQiw05nSFI8.

Thanks

• Thanks to Sameh Khamis' help