# CroCo-Stereo

**Repository Path**: ykuo/cro-co-stereo

## Basic Information

- **Project Name**: CroCo-Stereo
- **Description**: 复现先进算法：CroCo-Stereo
复现代码来源于开源代码：naver/croco
复现论文：CroCo v2: Improved Cross-view Completion Pre-training for Stereo Matching and Optical Flow
- **Primary Language**: Python
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2023-12-25
- **Last Updated**: 2025-08-16

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# CroCo-Stereo

#### 介绍

- 复现先进算法：CroCo-Stereo
- 复现代码来源于开源代码：[naver/croco](https://github.com/naver/croco)
- 复现论文：CroCo v2: Improved Cross-view Completion Pre-training for Stereo Matching and Optical Flow

#### 软件架构

- python3.11
- pytorch2.1.0+cu121

#### 安装教程

1.  xxxx
2.  xxxx
3.  xxxx

#### 使用说明

1.  tensorboard --logdir=./checkpoints/* --port=6006
2.  watch -n 1 nvidia-smi
3.  netstat -ano | findstr :6006
4.  nohup python ./main.py > ./checkpoints/*/nohup.log 2>&1 &
5.  tail -f ./checkpoints/*/nohup.log
6.  ps aux|grep python

#### 微调Finetune
    # To finetune on KITTI mixed 2012 and 2015, run
    python -u stereoflow/train.py stereo --crop 320 1184 --criterion "LaplacianLossBounded2()" --dataset "Kitti12('train')+Kitti15('train')" --val_dataset "Kitti15('test')" --batch_size 2 --num_workers 10 --epochs 20 --start_from stereoflow_models/crocostereo.pth --output_dir checkpoints/finetune_kitti/
    # To finetune on USVInland, run
    python -u stereoflow/train.py stereo --crop 320 640 --criterion "LaplacianLossBounded2()" --dataset "USVInland('train')" --val_dataset "USVInland('test')" --batch_size 4 --num_workers 10 --epochs 20  --start_from stereoflow_models/crocostereo.pth --output_dir checkpoints/finetune_usvinland/
    # To finetune on USVInland_seg, run
    python -u stereoflow/train.py stereo --crop 320 640 --criterion "LaplacianLossBounded2()" --dataset "USVInland_seg('train')" --val_dataset "USVInland_seg('test')" --batch_size 4 --num_workers 10 --epochs 20 --start_from stereoflow_models/crocostereo.pth --output_dir checkpoints/finetune_usvinland_seg/

#### 测试Test
    # To test on KITTI 2015, run
    python stereoflow/test.py --model stereoflow_models/crocostereo_finetune_kitti.pth --dataset "Kitti15('test')" --save submission --tile_overlap 0.9
    # To test on USVInland, run
    python stereoflow/test.py --model stereoflow_models/crocostereo_finetune_usvinland.pth --dataset "USVInland('test')" --save submission --tile_overlap 0.9
    # To test on USVInland_seg, run
    python stereoflow/test.py --model stereoflow_models/crocostereo_finetune_usvinland_seg.pth --dataset "USVInland_seg('test')" --save submission --tile_overlap 0.9

#### 演示Demo
    # Input dataset, imgname(without suffix), out in demo folder with gt and pred(and norm) in txt, pred pseudo img and prob histogram, default kitti run
    python stereoflow/croco-stereo-demo.py --dataset kitti --imgname 000000_10
    # Demo for usvinland(or --dataset usvinland_seg), run
    python stereoflow/croco-stereo-demo.py --dataset usvinland --imgname H05_1_0000000600

#### 参与贡献

1.  Fork 本仓库
2.  新建 Feat_xxx 分支
3.  提交代码
4.  新建 Pull Request

#### 特技

1.  使用 Readme\_XXX.md 来支持不同的语言，例如 Readme\_en.md, Readme\_zh.md
2.  Gitee 官方博客 [blog.gitee.com](https://blog.gitee.com)
3.  你可以 [https://gitee.com/explore](https://gitee.com/explore) 这个地址来了解 Gitee 上的优秀开源项目
4.  [GVP](https://gitee.com/gvp) 全称是 Gitee 最有价值开源项目，是综合评定出的优秀开源项目
5.  Gitee 官方提供的使用手册 [https://gitee.com/help](https://gitee.com/help)
6.  Gitee 封面人物是一档用来展示 Gitee 会员风采的栏目 [https://gitee.com/gitee-stars/](https://gitee.com/gitee-stars/)

#### 引用
> @inproceedings{croco,
  title={{CroCo: Self-Supervised Pre-training for 3D Vision Tasks by Cross-View Completion}},
  author={{Weinzaepfel, Philippe and Leroy, Vincent and Lucas, Thomas and Br\'egier, Romain and Cabon, Yohann and Arora, Vaibhav and Antsfeld, Leonid and Chidlovskii, Boris and Csurka, Gabriela and Revaud J\'er\^ome}},
  booktitle={{NeurIPS}},
  year={2022}
}

> @inproceedings{croco_v2,
  title={{CroCo v2: Improved Cross-view Completion Pre-training for Stereo Matching and Optical Flow}},
  author={Weinzaepfel, Philippe and Lucas, Thomas and Leroy, Vincent and Cabon, Yohann and Arora, Vaibhav and Br{\'e}gier, Romain and Csurka, Gabriela and Antsfeld, Leonid and Chidlovskii, Boris and Revaud, J{\'e}r{\^o}me},
  booktitle={ICCV},
  year={2023}
}