• Search by Module
• Search by Words
• Search Projects

• Java
• Python
• JavaScript
• TypeScript
• C++
• Scala
• Blog

Project: DSMnet (GitHub Link)

• DSMnet-master
  • myTransforms
    • aug_spatial.py
    • aug_color-t1.py
    • aug_color.py
    • Color.py
    • __init__.py
  • DSMnet_finetune.sh
  • utils
    • evaluate.py
    • imwrap.py
    • __init__.py
    • utils.py
  • LICENSE
  • models
    • dispnet.py
    • iresnet.py
    • psmnet
      • basic.py
      • stackhourglass.py
      • __init__.py
      • submodule.py
    • util_fun.py
    • test_models_time.py
    • util_conv.py
    • __init__.py
    • test_models_memory.py
    • dispnetcorr.py
    • gcnet.py
    • test_models_debug.py
  • DSMnet_test.sh
  • deploy
    • deploy.py
    • deploy.sh
  • losses
    • test_loss.py
    • SSIM.py
    • SSIM0.py
    • loss.py
    • __init__.py
    • test_ssim.py
  • debug_dataloader_stereo.py
  • stereo.py
  • DSMnet_train_sceneflow.sh
  • DSMnet_train_kitti-raw.sh
  • main.py
  • README.md
  • stereo_selfsupervised.py
  • myDatasets_stereo
    • Dataset_stereo.py
    • stereo_check.py
    • stereo_paths.py
    • test_stereo_paths.py
    • __init__.py
    • img_rw.py
    • test_stereo_check.py
    • img_rw_pfm.py
  • DSMnet_submit.sh
  • DSMnet_train_kitti.sh
  • stereo_supervised.py

Pytorch implementation of the several Deep Stereo Matching Network(DSMnet)

Contents

1. Introduction
2. Usage
3. Results
4. Licensing

Introduction

Pytorch implementation of the several Deep Stereo Matching Network

Supported models:

• Dispnet
• DispnetC
• GCnet
• iResnet
• PSMnet

Usage

Dependencies

• Python2.7
• PyTorch(0.3.0+)
• KITTI Stereo
• Scene Flow

Train

As an example, use the following command to train a DSM(such as DispnetCorr) on a dataset(such as KITTI2015)

./DSMnet_train_kitti.sh

As another example, use the following command to finetune a DSM(such as DispnetCorr) on a dataset(such as KITTI2015)

./DSMnet_finetune.sh

You need see the files(train_dispnetcorr_kitti.sh and train_dispnetcorr_kitti.sh) for details. You can alse change the DSM or dataset for train or finetune in the files.

submit

Use the following command to evaluate the trained DSM(such as DispnetCorr) on a dataset(such as KITTI2015 test data) without ground truth.

./DSMnet_submit.sh

You need see the file(submit_dispnetcorr_kitti.sh) for details.

Pretrained Model

KITTI	KITTI-raw-ss
DispnetC-pretrained-kitti	DispnetC-pretrained-kitti-raw-ss

Results

Qualitative results

Left image

Right image

Predicted disparity

Licensing

Unless otherwise stated, the source code and trained Torch and Python model files are copyright Carnegie Mellon University and licensed under the Apache 2.0 License. Portions from the following third party sources have been modified and are included in this repository. These portions are noted in the source files and are copyright their respective authors with the licenses listed.