details.txt

1
Register Team
Register or manage your team to receive a unique team ID, updates, and access to submission instructions.



Register or manage team
2
Receive Unique Team ID
After registration, each team is issued a unique identifier that must be used for all benchmark submissions and leaderboard entries. Include this team number and your exact registered team name in the abstract of your workshop paper.


these are my details Team ID
RoCo-45
Team name
Flow State
Primary contact/login email
syedmuhaimintahsin@gmail.com
Created
Sep 8, 2026, 10:32 PM
Last updated
Sep 8, 2026, 10:45 PM
Revision
2
Selected Tracks
Optical Flow
Stereo Matching
Scene Flow
Exploration Track
Organizer Record Sync
Organizer spreadsheet synchronized at revision 2.

Team Members
Name	Email	Affiliation
Syed Mohaiminul Hoque	syedmuhaimintahsin@gmail.com	Center For Computational and Data Sciences Lab, Independent University Bangladesh
3
Download Data
Download the Spring (https://darus.uni-stuttgart.de/dataset.xhtml?persistentId=doi:10.18419/darus-3376) and RobustSpring (https://darus.uni-stuttgart.de/dataset.xhtml?persistentId=doi:10.18419/DARUS-5047) training splits needed for model development, local validation, and practice runs.

4
Install the Starter Kit
(https://github.com/hmorimitsu/roco-spring-devkit)
Set up the PTLFlow-based Starter Kit with training and inference scripts, data loaders, and example configurations.

GitHub Starter Kit
5
Run Baseline
Run the provided baseline configurations to verify your environment, data paths, and end-to-end inference pipeline.

6
Generate Predictions
Produce model outputs for your chosen task(s) and export them in the required prediction format.

7
Validate Archive Locally
Package your predictions into a submission archive and run local validation to catch formatting or naming issues before uploading.

8
Submit to Benchmark Server
Upload your validated archive to the benchmark(https://spring-benchmark.org/) to receive scores on the development leaderboard.
(Spring Dataset and Benchmark Logo
Spring: L. Mehl, J. Schmalfuss, A. Jahedi, Y. Nalivayko, A. Bruhn — University of Stuttgart

RobustSpring: V. Oei, J. Schmalfuss, L. Mehl, M. Bartsch, S. Agnihotri, M. Keuper, A. Bulling, A. Bruhn — University of Stuttgart, University of Mannheim, MPI for Informatics

Download Stereo Optical Flow Scene Flow Submit FAQ
syedmuhaimintahsin@gmail.com | Logout
Submission
Create New Submission
✖ Your account is not verified yet by our team. Please check again later.

Instructions
To participate in the benchmark, prepare and submit your method as follows. You can optionally include robustness evaluation results alongside the standard evaluation.

Prepare predictions on the Spring test split.
Use our Python I/O utilities for .flo5 and .dsp5: flow_IO.py.
Follow the exact folder structure, file naming conventions, and formats used in the dataset.
Standard evaluation folder structure:
Stereo: <rootdir>/####/disp1_{left|right}/disp1_{left|right}_####.dsp5
Optical Flow: <rootdir>/####/flow_{FW|BW}_{left|right}/flow_{FW|BW}_{left|right}_####.flo5
Scene Flow (add disparity over time): <rootdir>/####/disp2_{FW|BW}_{left|right}/disp2_{FW|BW}_{left|right}_####.dsp5
Optional: Robustness evaluation
Submit predictions on corrupted versions of the test set (e.g., fog, noise, blur).
Use the same naming conventions, with an extra top-level folder per corruption.
Robustness evaluation folder structure:
Stereo: <rootdir>/<corruption>/test/####/disp1_{left|right}/disp1_{left|right}_####.dsp5
Optical Flow: <rootdir>/<corruption>/test/####/flow_{FW|BW}_{left|right}/flow_{FW|BW}_{left|right}_####.flo5
Scene Flow: <rootdir>/<corruption>/test/####/disp2_{FW|BW}_{left|right}/disp2_{FW|BW}_{left|right}_####.dsp5
Corruption folder names: clean, brightness, contrast, defocus_blur, elastic_transform, fog, frost, gaussian_blur, gaussian_noise, glass_blur, impulse_noise, jpeg_compression, motion_blur, pixelate, rain, saturate, shot_noise, snow, spatter, speckle_noise, zoom_blur.
Generate submission file(s) with subsampling tools.
Download the executables: subsampling tools.
Run in your rootdir:
./disp1_subsampling <rootdir> — stereo
./flow_subsampling <rootdir> — optical flow
./disp2_subsampling <rootdir> — scene flow (with the other two)
For robustness evaluation, run the corresponding “robust” executables in your root directory containing the corruption subfolders:
./disp1_robust_subsampling <rootdir> for stereo robustness
./flow_robust_subsampling <rootdir> for optical flow robustness
./disp2_robust_subsampling <rootdir> for scene flow robustness (alongside the other two)
Each command generates one .hdf5 file in the current directory, ready for upload.
Submit results on the benchmark platform.
Click the “Create new submission” button above and upload your .hdf5 file(s).
For scene flow, upload all three files. You may also upload additional robustness evaluation files under the same method entry.
Evaluation takes ~1-2 hours. You'll receive a notification when done.
Results default to private. Later you can switch visibility to:
Private (only you)
Public Anonymous (no author info)
Public (with author and method name)
Your Submissions
No data available.

Lukas Mehl, Jenny Schmalfuss, Azin Jahedi, Yaroslava Nalivayko, Andrés Bruhn:
"Spring: A High-Resolution High-Detail Dataset and Benchmark for Scene Flow, Optical Flow and Stereo".
In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023.

Victor Oei, Jenny Schmalfuss, Lukas Mehl, Madlen Bartsch, Shashank Agnihotri, Margret Keuper, Andreas Bulling, Andrés Bruhn:
"RobustSpring: Benchmarking Robustness to Image Corruptions for Optical Flow, Scene Flow and Stereo".
In The Fourteenth International Conference on Learning Representations (ICLR), 2026.

Dog from the spring movie called autumn	Logo of University of Stuttgart / VIS Institute	Logo of SFB TRR 161	Logo of IMPRS-IS
The Spring movie assets by Blender Foundation are licensed under CC BY 4.0. The authors thank Andy Goralczyk, Francesco Siddi and the whole Blender Studio team. Lukas Mehl, Jenny Schmalfuss, and Victor Oei acknowledge funding by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project-ID 251654672 – TRR 161: Quantitative Methods for Visual Computing (B04, A07). Jenny Schmalfuss and Andres Bruhn acknowledge support by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project-ID 533085500 – Robust Optical Flow. Shashank Agnihotri and Margret Keuper acknowledge support by the DFG Research Unit 5336 – Learning to Sense (L2S). Finally, Jenny Schmalfuss, Azin Jahedi, and Victor Oei acknowledge support from the International Max Planck Research School for Intelligent Systems (IMPRS-IS).

Contact | Imprint | Privacy)


Spring Dataset and Benchmark Logo
Spring: L. Mehl, J. Schmalfuss, A. Jahedi, Y. Nalivayko, A. Bruhn — University of Stuttgart

RobustSpring: V. Oei, J. Schmalfuss, L. Mehl, M. Bartsch, S. Agnihotri, M. Keuper, A. Bulling, A. Bruhn — University of Stuttgart, University of Mannheim, MPI for Informatics

Download Stereo Optical Flow Scene Flow Submit FAQ
syedmuhaimintahsin@gmail.com | Logout
Download Datasets
Our benchmark has two dataset components:

Standard evaluation (required): Download the Standard Spring dataset below to measure accuracy on the clean test split.
Robustness evaluation (optional): Download the Spring robustness dataset if you also want to evaluate performance under corruptions.
Standard Spring dataset
Provides clean test-split ground truth for stereo, optical flow, and scene flow on 10 driving sequences.
Download Standard Spring dataset
(DOI: 10.18419/DARUS-3376)
Spring robustness dataset
Contains the same sequences under 21 realistic corruptions (fog, noise, blur, etc.) for robustness evaluation.
Download Spring robustness dataset
(DOI: 10.18419/DARUS-5047)
After downloading, extract each archive into your chosen rootdir. Your folder tree will include entries like 0003/flow_FW_left/… for the standard split and <corruption>/0003/flow_FW_left/… for robustness. See the FAQ for full instructions on folder structure and submission.

Lukas Mehl, Jenny Schmalfuss, Azin Jahedi, Yaroslava Nalivayko, Andrés Bruhn:
"Spring: A High-Resolution High-Detail Dataset and Benchmark for Scene Flow, Optical Flow and Stereo".
In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023.

Victor Oei, Jenny Schmalfuss, Lukas Mehl, Madlen Bartsch, Shashank Agnihotri, Margret Keuper, Andreas Bulling, Andrés Bruhn:
"RobustSpring: Benchmarking Robustness to Image Corruptions for Optical Flow, Scene Flow and Stereo".
In The Fourteenth International Conference on Learning Representations (ICLR), 2026.

Dog from the spring movie called autumn	Logo of University of Stuttgart / VIS Institute	Logo of SFB TRR 161	Logo of IMPRS-IS
The Spring movie assets by Blender Foundation are licensed under CC BY 4.0. The authors thank Andy Goralczyk, Francesco Siddi and the whole Blender Studio team. Lukas Mehl, Jenny Schmalfuss, and Victor Oei acknowledge funding by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project-ID 251654672 – TRR 161: Quantitative Methods for Visual Computing (B04, A07). Jenny Schmalfuss and Andres Bruhn acknowledge support by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project-ID 533085500 – Robust Optical Flow. Shashank Agnihotri and Margret Keuper acknowledge support by the DFG Research Unit 5336 – Learning to Sense (L2S). Finally, Jenny Schmalfuss, Azin Jahedi, and Victor Oei acknowledge support from the International Max Planck Research School for Intelligent Systems (IMPRS-IS).

Contact | Imprint | Privacy

Skip to main content
DaRUS homepage
Search 
About
User Guide
Support
Log In
SFB-TRR 161 B04 "Adaptive Algorithms for Motion Estimation" logo	
SFB-TRR 161 B04 "Adaptive Algorithms for Motion Estimation"
(Universität Stuttgart)
DaRUS>SFB/Transregio 161 "Quantitative Methods for Visual Computing">SFB-TRR 161 B04 "Adaptive Algorithms for Motion Estimation">
RobustSpring: Benchmarking Robustness to Image Corruptions for Optical Flow, Scene Flow and Stereo
Version 1.0
Schmalfuss, Jenny; Oei, Victor; Mehl, Lukas; Bartsch, Madlen; Agnihotri, Shashank; Keuper, Margret; Bruhn, Andres, 2025, "RobustSpring: Benchmarking Robustness to Image Corruptions for Optical Flow, Scene Flow and Stereo", https://doi.org/10.18419/DARUS-5047, DaRUS, V1
Learn about Data Citation Standards.
Dataset Metrics 
35,314 Downloads 
Description 	
The RobustSpring dataset contains the image corruption data files for scene flow, optical flow and stereo estimation with the Spring dataset. Note that this repository contains only the Spring test data files. For easier handling, we organized them into sub-directories by image corruption type:
brightness.zip : brightness image corruption
contrast.zip : contrast image corruption
defocus_blur.zip : defocus_blur image corruption
elastic_transform.zip : elastic_transform image corruption
fog.zip : fog image corruption
frost.zip : frost image corruption
gaussian_blur.zip : gaussian_blur image corruption
gaussian_noise.zip : gaussian_noise image corruption
glass_blur.zip : glass_blur image corruption
impulse_noise.zip : impulse_noise image corruption
jpeg_compression.zip : jpeg_compression image corruption
motion_blur.zip : motion_blur image corruption
pixelate.zip : pixelate image corruption
rain.zip : rain image corruption
saturate.zip : saturate image corruption
shot_noise.zip : shot_noise image corruption
snow.zip : snow image corruption
spatter.zip : spatter image corruption
speckle_noise.zip : speckle_noise image corruption
zoom_blur.zip : zoom_blur image corruption

Each image corruption folder is internally organized as follows:
test : Indicates that this is the test proportion of the Spring dataset
0003: Scene subfolder for scene 0003, 131 frames
frame_left: Left frames
frame_left_0001.png: Frame 0001
...
frame_left_0131.png: Frame 0131
frame_right: Right frames
frame_right_0001.png: Frame 0001
...
frame_right_0131.png: Frame 0131
0019: Scene 0019, same internal structure as above, 111 frames.
0028: Scene 0028, same internal structure as above, 39 frames.
0028: Scene 0029, same internal structure as above, 135 frames.
0031: Scene 0031, same internal structure as above, 73 frames.
0034: Scene 0034, same internal structure as above, 47 frames.
0035: Scene 0035, same internal structure as above, 120 frames.
0040: Scene 0040, same internal structure as above, 111 frames.
0042: Scene 0042, same internal structure as above, 116 frames.
0046: Scene 0046, same internal structure as above, 117 frames.

File Formats:
All images are given as .png files

For the project website see spring-benchmark.org. (2025-05-13)
Subject 	Computer and Information Science
Keyword 	Computer Vision, Optical Flow, Computer Stereo Vision, Benchmark
Related Publication 	Jenny Schmalfuss, Victor Oei, Lukas Mehl, Madlen Bartsch, Shashank Agnihotri, Margret Keuper, Andrés Bruhn: RobustSpring: Benchmarking Robustness to Image Corruptions for Optical Flow, Scene Flow and Stereo. 2025arXiv: abs/2505.09368
Data Generation 	Simulation
License/Data Use Agreement	
Creative Commons Attribution 4.0 International License. CC BY 4.0

FilesMetadataTermsVersions
Change View

Table

Tree
Search
Search this dataset...
Filter by
1 to 20 of 20 Files
brightness.zip
Spring-corrupted-submission-zips/
ZIP Archive - 6.1 GB
Published May 14, 2025
1,830 Downloads
MD5: fd3...495 
contrast.zip
Spring-corrupted-submission-zips/
ZIP Archive - 2.8 GB
Published May 14, 2025
1,684 Downloads
MD5: 9df...0a0 
defocus_blur.zip
Spring-corrupted-submission-zips/
ZIP Archive - 2.7 GB
Published May 14, 2025
1,715 Downloads
MD5: 0a6...921 
elastic_transform.zip
Spring-corrupted-submission-zips/
ZIP Archive - 5.0 GB
Published May 14, 2025
1,713 Downloads
MD5: a76...f04 
fog.zip
Spring-corrupted-submission-zips/
ZIP Archive - 4.1 GB
Published May 14, 2025
1,666 Downloads
MD5: 938...9df 
frost.zip
Spring-corrupted-submission-zips/
ZIP Archive - 6.1 GB
Published May 14, 2025
2,072 Downloads
MD5: ef5...01c 
gaussian_blur.zip
Spring-corrupted-submission-zips/
ZIP Archive - 2.4 GB
Published May 14, 2025
1,793 Downloads
MD5: 55f...e37 
gaussian_noise.zip
Spring-corrupted-submission-zips/
ZIP Archive - 10.7 GB
Published May 14, 2025
1,716 Downloads
MD5: a15...176 
glass_blur.zip
Spring-corrupted-submission-zips/
ZIP Archive - 3.1 GB
Published May 14, 2025
1,590 Downloads
MD5: 430...898 
impulse_noise.zip
Spring-corrupted-submission-zips/
ZIP Archive - 7.3 GB
Published May 14, 2025
1,622 Downloads
MD5: 27b...c04 
jpeg_compression.zip
Spring-corrupted-submission-zips/
ZIP Archive - 1.0 GB
Published May 14, 2025
1,635 Downloads
MD5: 865...eb0 
motion_blur.zip
Spring-corrupted-submission-zips/
ZIP Archive - 4.6 GB
Published May 14, 2025
1,755 Downloads
MD5: bbe...357 
pixelate.zip
Spring-corrupted-submission-zips/
ZIP Archive - 306.3 MB
Published May 14, 2025
1,665 Downloads
MD5: 309...9a8 
rain.zip
Spring-corrupted-submission-zips/
ZIP Archive - 6.4 GB
Published May 14, 2025
1,963 Downloads
MD5: e8b...070 
saturate.zip
Spring-corrupted-submission-zips/
ZIP Archive - 5.4 GB
Published May 14, 2025
1,628 Downloads
MD5: aa9...ebc 
shot_noise.zip
Spring-corrupted-submission-zips/
ZIP Archive - 6.5 GB
Published May 14, 2025
1,590 Downloads
MD5: 919...7c4 
snow.zip
Spring-corrupted-submission-zips/
ZIP Archive - 6.5 GB
Published May 14, 2025
1,741 Downloads
MD5: 509...cd8 
spatter.zip
Spring-corrupted-submission-zips/
ZIP Archive - 6.5 GB
Published May 14, 2025
1,802 Downloads
MD5: c3d...4a6 
speckle_noise.zip
Spring-corrupted-submission-zips/
ZIP Archive - 10.8 GB
Published May 14, 2025
1,748 Downloads
MD5: 7d2...833 
zoom_blur.zip
Spring-corrupted-submission-zips/
ZIP Archive - 3.5 GB
Published May 14, 2025
2,386 Downloads
MD5: 5cb...27c 
F
P
1
N
E
Files Per Page 
50
Copyright © 2026 | Imprint | Terms of Use | Privacy Policy

Powered by The Dataverse Project logov. 6.11
 
Skip to main content
DaRUS homepage
Search 
About
User Guide
Support
Log In
SFB-TRR 161 B04 "Adaptive Algorithms for Motion Estimation" logo	
SFB-TRR 161 B04 "Adaptive Algorithms for Motion Estimation"
(Universität Stuttgart)
DaRUS>SFB/Transregio 161 "Quantitative Methods for Visual Computing">SFB-TRR 161 B04 "Adaptive Algorithms for Motion Estimation">
Spring: A High-Resolution High-Detail Dataset and Benchmark for Scene Flow, Optical Flow and Stereo
Version 2.0
Mehl, Lukas; Schmalfuss, Jenny; Jahedi, Azin; Nalivayko, Yaroslava; Bruhn, Andrés, 2023, "Spring: A High-Resolution High-Detail Dataset and Benchmark for Scene Flow, Optical Flow and Stereo", https://doi.org/10.18419/DARUS-3376, DaRUS, V2
Learn about Data Citation Standards.
Dataset Metrics 
378,947 Downloads 
Description 	
The Spring dataset contains files for scene flow, optical flow and stereo estimation. For easier handling, we organized them into sub-directories:
train split:
train_frame_left.zip: left camera frames
train_frame_right.zip: right camera frames
train_disp1_left.zip: left-to-right disparity in the reference frame
train_disp1_right.zip: right-to-left disparity in the reference frame
train_disp2_FW_left.zip: left-to-right disparity in the future/forward target frame
train_disp2_BW_left.zip: left-to-right disparity in the past/backward target frame
train_disp2_FW_right.zip: right-to-left disparity in the future/forward target frame
train_disp2_BW_right.zip: right-to-left disparity in the past/backward target frame
train_flow_FW_left.zip: left forward optical flow
train_flow_BW_left.zip: left backward optical flow
train_flow_FW_right.zip: right forward optical flow
train_flow_BW_right.zip: right backward optical flow
train_cam_data.zip: camera data: intrinsics, extrinsics, focal distance
train_maps.zip: additional maps: detail, match, rigid, sky
test split:
test_frame_left.zip: left camera frames
test_frame_right.zip: right camera frames
test_cam_data.zip: camera data: intrinsics

File formats:
images and maps are given in png format
optical flow files are given in HDF5 file format and named .flo5
disparity files are given in HDF5 file format and named .dsp5
For the project website see spring-benchmark.org.
Subject 	Computer and Information Science
Keyword 	Computer Vision, Optical Flow, Computer Stereo Vision, Benchmark
Related Publication 	Lukas Mehl, Jenny Schmalfuss, Azin Jahedi, Yaroslava Nalivayko, Andrés Bruhn: Spring: A High-Resolution High-Detail Dataset and Benchmark for Scene Flow, Optical Flow and Stereo. Proc. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023.arXiv: 2303.01943
License/Data Use Agreement	
Creative Commons Attribution 4.0 International License. CC BY 4.0

FilesMetadataTermsVersions
Search
Search this dataset...
1 to 17 of 17 Files
test_cam_data.zip
ZIP Archive - 3.7 KB
Published Mar 14, 2023
1,658 Downloads
MD5: 7de...ecd 
test_frame_left.zip
ZIP Archive - 2.8 GB
Published Mar 14, 2023
6,951 Downloads
MD5: f0c...4b4 
test_frame_right.zip
ZIP Archive - 2.8 GB
Published Mar 14, 2023
11,191 Downloads
MD5: 914...12d 
train_cam_data.zip
ZIP Archive - 466.9 KB
Published Mar 14, 2023
1,915 Downloads
MD5: 9b1...d94 
train_disp1_left.zip
ZIP Archive - 9.9 GB
Published Mar 14, 2023
43,560 Downloads
MD5: 953...257 
train_disp1_right.zip
ZIP Archive - 9.8 GB
Published Mar 14, 2023
32,024 Downloads
MD5: 867...910 
train_disp2_BW_left.zip
ZIP Archive - 9.8 GB
Published Mar 14, 2023
3,958 Downloads
MD5: 32e...6e4 
train_disp2_BW_right.zip
ZIP Archive - 9.8 GB
Published Mar 14, 2023
3,249 Downloads
MD5: 703...b13 
train_disp2_FW_left.zip
ZIP Archive - 9.8 GB
Published Mar 14, 2023
3,955 Downloads
MD5: 1e3...feb 
train_disp2_FW_right.zip
ZIP Archive - 9.8 GB
Published Jul 17, 2023
3,089 Downloads
MD5: bf5...fcb 
train_flow_BW_left.zip
ZIP Archive - 44.0 GB
Published Mar 14, 2023
12,355 Downloads
MD5: da6...fcc 
train_flow_BW_right.zip
ZIP Archive - 43.7 GB
Published Mar 14, 2023
10,844 Downloads
MD5: 551...881 
train_flow_FW_left.zip
ZIP Archive - 44.0 GB
Published Mar 14, 2023
55,968 Downloads
MD5: 5cf...664 
train_flow_FW_right.zip
ZIP Archive - 43.7 GB
Published Jul 17, 2023
46,218 Downloads
MD5: d19...5c7 
train_frame_left.zip
ZIP Archive - 12.4 GB
Published Mar 14, 2023
51,460 Downloads
MD5: 3a5...71a 
train_frame_right.zip
ZIP Archive - 12.4 GB
Published Mar 14, 2023
35,642 Downloads
MD5: 3b6...936 
train_maps.zip
ZIP Archive - 4.3 GB
Published Jul 17, 2023
54,742 Downloads
MD5: 6f6...e48 
F
P
1
N
E
Files Per Page 
50
Copyright © 2026 | Imprint | Terms of Use | Privacy Policy

Powered by The Dataverse Project logov. 6.11