Home
Uni-Logo
 

Datasets

Binaries/Code Datasets Open Source Software

Freiburg-Berkeley Motion Segmentation Dataset
Video Segmentation Benchmark
Image Sequences
TEM Dataset
TILDA Textile Texture Database
Training data for Exemplar CNN
Generated Matching Dataset
Training data for chair generation
Stereo Ego-motion Dataset
Optical Flow Datasets: "Flying Chairs", "ChairsSDHom"
Scene Flow Datasets
Human Part Segmentation Datasets  
Rendered Handpose Dataset
Pedestrian Zone Scene
FreiHAND Dataset
HanCo Dataset
Human Pose RGBD Datasets
OVAD: Open-Vocabulary Attribute Detection Dataset
ADE-OoD: a benchmark for dense Out-of-Distribution detection on natural images.


Freiburg-Berkeley Motion Segmentation Dataset (FBMS-59)



The original Berkeley Motion Segmentation Dataset (BMS-26) consists of 26 video sequences with pixel-accurate segmentation annotation of moving objects. A total of 189 frames is annotated. 12 of the sequences are taken from the Hopkins 155 dataset and new annotation is added.

The Freiburg-Berkeley Motion Segmentation Dataset (FBMS-59) is an extension of the BMS dataset with 33 additional video sequences. A total of 720 frames is annotated. FBMS-59 comes with a split into a training set and a test set. Typical challenges appear in both sets.

Both datasets come with evaluation software that allows direct comparison of results. The evaluation software is prepared for and has been tested with 64-bit Linux. If you use another operating system, you must modify the source code of this software to make it run on your computer. We highly recommend running the evaluation on a Linux system. For FBMS-59 the evaluation protocol and metrics have been improved. We recommend using the larger FBMS-59 dataset. BMS-26 is maintained for comparison to previous works.

See the readme file to learn how to use the dataset and the tools.


Terms of use

The datasets are provided only for research purposes and without any warranty. Any commercial use is prohibited. When using the BMS-26 or FBMS-59 in your research work, you should cite the following papers, respectively:

T. Brox, J. Malik
Object segmentation by long term analysis of point trajectories,
European Conference on Computer Vision (ECCV), September 2010.

P. Ochs, J. Malik, T. Brox
Segmentation of moving objects by long term video analysis,
IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014.


Download BMS-26 dataset and evaluation software (260MB) [md5sum: be88c66bfe37cf1ee851364f130bf0df]

Download FBMS-59 training set (504MB)
Download uncompressed FBMS-59 training set (2.6GB, slightly higher image quality)
Download FBMS-59 test set (386MB)
Download uncompressed FBMS-59 test set (2.1GB, slightly higher image quality)
Download FBMS-59 evaluation software

There is additional ground truth annotation for 3D motion segmentation.