Data

Tools

Indexes

Applications

Experiments

Awesome

An open API service indexing awesome lists of open source software.

https://github.com/div99/w-stereo-disp

(NeurIPS 2020 Spotlight) Wasserstein Distances for Stereo Disparity Estimation
https://github.com/div99/w-stereo-disp

Last synced: 10 months ago
JSON representation

(NeurIPS 2020 Spotlight) Wasserstein Distances for Stereo Disparity Estimation

Awesome Lists containing this project

README 

          
# Wasserstein Distances for Stereo Disparity Estimation



[**[Project Page](https://div99.github.io/W-Stereo-Disp)**] Accepted in NeurIPS 2020 as Spotlight



[Wasserstein Distances for Stereo Disparity Estimation](https://arxiv.org/abs/2007.03085) 



by [Divyansh Garg](https://divyanshgarg.com), [Yan Wang](https://www.cs.cornell.edu/~yanwang/), [Bharath Hariharan](http://home.bharathh.info/), [Mark Campbell](https://campbell.mae.cornell.edu/), [Kilian Q. Weinberger](http://kilian.cs.cornell.edu/) and [Wei-Lun Chao](http://www-scf.usc.edu/~weilunc/)



![Figure](figures/neurips2020-pipeline.png)



### Citation

```

@inproceedings{div2020wstereo,

  title={Wasserstein Distances for Stereo Disparity Estimation},

  author={Garg, Divyansh and Wang, Yan and Hariharan, Bharath and Campbell, Mark and Weinberger, Kilian and Chao, Wei-Lun},

  booktitle={NeurIPS},

  year={2020}

}

```



## Introduction

Existing approaches to depth or disparity estimation output a distribution over a

set of pre-defined discrete values. This leads to inaccurate results when the true

depth or disparity does not match any of these values. The fact that this distribution

is usually learned indirectly through a regression loss causes further problems in

ambiguous regions around object boundaries. We address these issues using a new

neural network architecture that is capable of outputting arbitrary depth values,

and a new loss function that is derived from the Wasserstein distance between

the true and the predicted distributions. We validate our approach on a variety

of tasks, including stereo disparity and depth estimation, and the downstream 3D

object detection. Our approach drastically reduces the error in ambiguous regions,

especially around object boundaries that greatly affect the localization of objects in

3D, achieving the state-of-the-art in 3D object detection for autonomous driving.



## Update

  - 1 June 2021: Released all pretrained models. Added script to download KITTI Object Detection dataset.



## Contents



Our Wasserstein loss [W_loss](https://github.com/Div99/W-Stereo-Disp/blob/85cbf2bdc199a632ffe6b960f91bdc8e02046b77/src/main_disp.py#L123) can be easily plugged in existing stereo depth models to improve the training and obtain better results.



We release the code for CDN-PSMNet and CDN-SDN models. 



- [Requirements](#requirements)

- [Pretrained Models](#pretrained-models)

- [Training and Inference](#training-and-inference)



## Requirements

1. Python 3.7

2. Pytorch 1.2.0+

3. CUDA

4. `pip install -r ./requirements.txt`

5. [SceneFlow](https://lmb.informatik.uni-freiburg.de/resources/datasets/SceneFlowDatasets.en.html)

5. [KITTI](http://www.cvlibs.net/datasets/kitti/eval_object.php?obj_benchmark=3d)



## Pretrained Models



Place the checkpoint folders in `./results`.



Depth Models

- [Trained on Sceneflow](https://drive.google.com/drive/folders/1uav_MD0cJ3jaeqkfJVwAv37r_cS_sKDn?usp=sharing)

- [Trained on KITTI Object Detection train set](https://drive.google.com/drive/folders/1c-YUzbuaS5wbBAmSznrP34Kn-lsWW56-?usp=sharing)

- [Trained on KITTI Object Detection train+val set](https://drive.google.com/drive/folders/1gePafBBvHJDm1b4EpTa34C3XqoPOz757?usp=sharing)



Disparity Models

- [Trained on Sceneflow](https://drive.google.com/drive/folders/1E0QxohXyuGad96P64QwxQc7PVL99GGJU?usp=sharing)

- [Trained on KITTI 2015 train set](https://drive.google.com/drive/folders/1B-Jpr6w8-IaFpw54vji2oxVh8PLtYhJo?usp=sharing)

## Datasets

You have to download the [SceneFlow](https://lmb.informatik.uni-freiburg.de/resources/datasets/SceneFlowDatasets.en.html) and [KITTI](http://www.cvlibs.net/datasets/kitti/eval_object.php?obj_benchmark=3d) datasets. The structures of the datasets are shown in below. 



KITTI can be automatically downloaded using `./scripts/download_kitti.sh`

#### SceneFlow Dataset Structure

```

SceneFlow

    | monkaa

        | frames_cleanpass

        | disparity

    | driving

        | frames_cleanpass

        | disparity

    | flyingthings3d

        | frames_cleanpass 

        | disparity

```

#### KITTI Object Detection Dataset Structure

```

KITTI

    | training

        | calib

        | image_2

        | image_3

        | velodyne

    | testing

        | calib

        | image_2

        | image_3

```

Generate soft-links of SceneFlow Datasets. The results will be saved in `./sceneflow` folder. Please change to fakepath `path-to-SceneFlow` to the SceneFlow dataset location before running the script.

```bash

python sceneflow.py --path path-to-SceneFlow --force

```



Convert the KITTI velodyne ground truths to depth maps. Please change to fakepath `path-to-KITTI` to the SceneFlow dataset location before running the script.

```bash

python ./src/preprocess/generate_depth_map.py --data_path path-to-KITTI/ --split_file ./split/trainval.txt

```



Optionally download [KITTI2015](http://www.cvlibs.net/datasets/kitti/eval_scene_flow.php?benchmark=stereo) datasets for evaluating stereo disparity models.



## Training and Inference

- [1 Train CDN from Scratch on SceneFlow Dataset](#1-train-sDNet-from-scratch-on-sceneflow-dataset)

- [2 Train CDN on KITTI Dataset](#2-train-sdnet-on-kitti-dataset)

- [3 Generate Predictions](#3-generate-predictions)

- [4 Train 3D Detection with Pseudo-LiDAR](#7-train-3d-detection-with-pseudo-lidar)



We have provided all pretrained models [Pretrained Models](#pretrained-models). If you only want to generate the predictions, you can directly go to step [3](#3-generate-predictions). 



We use config files to simplify argument parsing. The default setting requires four gpus to train. You can use smaller batch sizes which are `btrain` and `bval`, if you don't have enough gpus. 



We provide code for both stereo disparity and stereo depth models.



We optionally use [Losswise](https://losswise.com) to visualize training metrics. An API key can be obtained and added to a config key to enable it.

#### 1 Train CDN-SDN from Scratch on SceneFlow Dataset

```bash

python ./src/main_depth.py -c src/configs/sceneflow_w1.config

```

The checkpoints are saved in `./results/stack_sceneflow_w1/`.



Follow same procedure to train stereo disparity model, but use `src/main_disp.py` and change to a disparity config.



#### 2 Train CDN-SDN on KITTI Dataset

```bash

python ./src/main_depth.py -c src/configs/kitti_w1.config \

    --pretrain ./results/sceneflow_w1/checkpoint.pth.tar --datapath  path-to-KITTI/training/

```

Before running, please change the fakepath `path-to-KITTI/` to the correct one. `--pretrain` is the path to the  pretrained model on SceneFlow. The training results are saved in `./results/kitti_w1_train`.



If you are working on evaluating CDN on KITTI testing set, you might want to train CDN on *training+validation* sets. The training results will be saved in `./results/sdn_kitti_trainval`.

```bash

python ./src/main_depth.py -c src/configs/kitti_w1.config \

    --pretrain ./results/sceneflow_w1/checkpoint.pth.tar \

    --datapath  path-to-KITTI/training/ --split_train ./split/trainval.txt \

    --save_path ./results/sdn_kitti_trainval

 ```

The disparity models can also be trained on KITTI2015 datasets using `src/kitti2015_w1_disp.config`.



#### 3 Generate Predictions

Please change the fakepath `path-to-KITTI`. Moreover, if you use the our provided checkpoint, please modify the value of `--resume` to the checkpoint location. 



* a. Using the model trained on KITTI training set, and generating predictions on training + validation sets.

```bash

python ./src/main_depth.py -c src/configs/kitti_w1.config \

    --resume ./results/sdn_kitti_train/checkpoint.pth.tar --datapath  path-to-KITTI/training/ \

    --data_list ./split/trainval.txt --generate_depth_map --data_tag trainval

``` 

The results will be saved in `./results/sdn_kitti_train/depth_maps_trainval/`.



* b. Using the model trained on KITTI training + validation set, and generating predictions on testing sets. You will use them when you want to submit your results to the leaderboard.



The results will be saved in `./results/sdn_kitti_trainval_set/depth_maps_trainval/`.

```bash

# testing sets

python ./src/main_depth.py -c src/configs/kitti_w1.config \

    --resume ./results/sdn_kitti_trainval/checkpoint.pth.tar --datapath  path-to-KITTI/testing/ \

    --data_list=./split/test.txt --generate_depth_map --data_tag test

``` 

The results will be saved in `./results/sdn_kitti_trainval/depth_maps_test/`.



#### 4 Train 3D Detection with Pseudo-LiDAR

For training 3D object detection models, follow step 4 and after in the Pseudo-LiDAR_V2 repo https://github.com/mileyan/Pseudo_Lidar_V2.



## Results



### Results on the Stereo Disparity

![Figure](figures/results_disp.png)



### 3D Object Detection Results on KITTI leader board

![Figure](figures/results_object3d.png)



## Questions

Please feel free to email us if you have any questions. 



Divyansh Garg ([dg595@cornell.edu](mailto:dg595@cornell.edu?subject=[GitHub]%20W-Stereo_Disp)),

Yan Wang ([yw763@cornell.edu](mailto:yw763@cornell.edu?subject=[GitHub]%20W-Stereo_Disp)),

Wei-Lun Chao ([weilunchao760414@gmail.com](mailto:weilunchao760414@gmail.com?subject=[GitHub]%20W-Stereo_Disp))