Datasets:

dronefreak
/

BDD100K-Drivable-Area-Segmentation

Tasks:

Languages:

Size:

Tags:

License:

Dataset card Files Files and versions

xet

Community

The Dataset Viewer has been disabled on this dataset.

BDD100K: Drivable Area Segmentation Dataset

BDD100K Drivable Area Segmentation Dataset Banner

Unofficial redistribution of BDD100K's drivable-area segmentation task (100K-image subset), under BDD100K's own data license (non-commercial/educational/research redistribution, with notice, explicitly permitted).

Disclaimer

This repository is not an official release of BDD100K.

BDD100K was created by Fisher Yu, Haofeng Chen, Xin Wang, Wenqi Xian, Yingying Chen, Fangchen Liu, Vashisht Madhavan, and Trevor Darrell at UC Berkeley (BAIR), who retain all copyright. This repository does not claim ownership of any images, annotations, or metadata.

This redistribution is sourced directly from BDD100K's official download (registration required at https://bdd-data.berkeley.edu/), not from any third-party mirror. Unlike the semantic segmentation task (a separate, smaller 10K-image subset), drivable-area segmentation pairs with the full 100K image set used for detection: the same 70,000 train + 10,000 val images that make up dronefreak/BDD100K, here paired with their drivable-area masks instead of (or alongside) the box labels.

This repository bundles its own copy of the paired images for self-containment, so no cross-repository joins are required to use it. If you already have the images_100k set from the detection repo, the images here are byte-identical.

Companion repositories cover BDD100K's other tasks: object detection, semantic segmentation, and the per-image attribute classification tasks — weather, scenario, and period/time-of-day.

Dataset Overview

BDD100K's drivable-area task provides a coarse, 3-class pixel mask for every image in the 100K set that has train/val labels released: 70,000 train + 10,000 validation images at 1280x720, each with a per-pixel drivable-area mask. The taxonomy (verified directly from the official bdd100k/label/label.py source) distinguishes:

direct (id 0) — the lane the ego vehicle is currently in / can drive into directly
alternative (id 1) — an alternative lane reachable with a lane change
background (id 2) — everything else (not drivable area)

In the color_labels visualization PNGs (verified directly against sample files in this repository), direct is rendered red (255, 0, 0), alternative is rendered blue (0, 0, 255), and background is black (0, 0, 0).

This is a simpler, purely-navigational signal than the 19-class semantic segmentation task — it only cares about where the car can drive, not what kind of object/surface is elsewhere in the scene.

No test split. BDD100K's official test images for this task (20,000 images) have no released masks (held out for the leaderboard) and are not included here.

Changes from the Official Release

No format conversion. The official release already ships exactly what's needed — JPEG images, single-channel (L mode) label-id-encoded PNG masks, and RGB color-visualization PNGs — so this repository redistributes them unmodified, just reorganized and sharded for Hugging Face.
Images bundled for self-containment. The paired images_100k train/val images are re-included here rather than requiring a join against the dronefreak/BDD100K repo, matching the self-contained pattern used by this project's other BDD100K dataset cards.
Sharded for Hub limits. Images and masks are sharded into shard_NNN/ subdirectories of up to 8,000 files each to stay under Hugging Face's practical per-directory file-count ceiling. This has no effect on the data itself — filenames and content are unchanged.
No image or mask pixel content was modified. No examples were added or removed relative to the official release.

Dataset Structure

<repo>/
├── README.md
├── bdd100k_drivable_banner.jpg
└── data/
    ├── images_100k/
    │   └── 100k/
    │       ├── train/   shard_000 … shard_008/   (JPEGs)
    │       └── val/     shard_000 … shard_001/   (JPEGs)
    └── drivable_maps/
        ├── labels/
        │   ├── train/   shard_000 … shard_008/   (label-id PNGs)
        │   └── val/     shard_000 … shard_001/   (label-id PNGs)
        └── color_labels/
            ├── train/   shard_000 … shard_008/   (RGB color-visualization PNGs)
            └── val/     shard_000 … shard_001/   (RGB color-visualization PNGs)

Images and masks are linked by filename stem (e.g. <id>.jpg in images_100k/100k/train/, <id>_drivable_id.png in drivable_maps/labels/train/, <id>_drivable_color.png in drivable_maps/color_labels/train/) — exactly the official BDD100K naming convention, unmodified. There is no metadata.jsonl linking them; this repository ships the raw, as-provided files rather than a datasets-library-ready config, so the Hub viewer is disabled for this repository (see YAML viewer: false).

Splits: train 70,000 images · validation 10,000 images (80,000 total — matches BDD100K's official drivable-area train+val counts exactly).

Dataset Sources

Original Paper

BDD100K: A Diverse Driving Dataset for Heterogeneous Multitask Learning

Fisher Yu, Haofeng Chen, Xin Wang, Wenqi Xian, Yingying Chen, Fangchen Liu, Vashisht Madhavan, Trevor Darrell

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020, pages 2633-2642. DOI: 10.1109/CVPR42600.2020.00271

Official Resources

Official Website: https://www.vis.xyz/bdd100k/
Official Download Portal: https://bdd-data.berkeley.edu/
Official Code Repository: https://github.com/bdd100k/bdd100k
Label Definitions (source of the class taxonomy above): https://github.com/bdd100k/bdd100k/blob/master/bdd100k/label/label.py

Attribution

All credit for collecting and annotating this dataset belongs entirely to the original BDD100K authors: Fisher Yu, Haofeng Chen, Xin Wang, Wenqi Xian, Yingying Chen, Fangchen Liu, Vashisht Madhavan, and Trevor Darrell, and UC Berkeley / BAIR.

This repository only reorganizes and shards the official files for Hugging Face compatibility; it does not modify, reinterpret, or take credit for the underlying imagery or annotations.

If you use this dataset in your research, please cite the original publication below.

License

BDD100K's code repository is BSD-3-Clause, but that license applies only to the code, not the data. The data and labels (downloaded from https://bdd-data.berkeley.edu/) carry their own license, reproduced here in full as required by its own terms:

Copyright ©2018. The Regents of the University of California (Regents). All Rights Reserved.

THIS SOFTWARE AND/OR DATA WAS DEPOSITED IN THE BAIR OPEN RESEARCH COMMONS REPOSITORY ON 1/1/2021

Permission to use, copy, modify, and distribute this software and its documentation for educational, research, and not-for-profit purposes, without fee and without a signed licensing agreement; and permission to use, copy, modify and distribute this software for commercial purposes (such rights not subject to transfer) to BDD and BAIR Commons members and their affiliates, is hereby granted, provided that the above copyright notice, this paragraph and the following two paragraphs appear in all copies, modifications, and distributions. Contact The Office of Technology Licensing, UC Berkeley, 2150 Shattuck Avenue, Suite 510, Berkeley, CA 94720-1620, (510) 643-7201, otl@berkeley.edu, http://ipira.berkeley.edu/industry-info for commercial licensing opportunities.

IN NO EVENT SHALL REGENTS BE LIABLE TO ANY PARTY FOR DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES, INCLUDING LOST PROFITS, ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN IF REGENTS HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

REGENTS SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE AND ACCOMPANYING DOCUMENTATION, IF ANY, PROVIDED HEREUNDER IS PROVIDED "AS IS". REGENTS HAS NO OBLIGATION TO PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.

Source: https://github.com/bdd100k/bdd100k/blob/master/doc/source/license.rst

Accordingly:

Redistribution for educational, research, and not-for-profit purposes is explicitly permitted, without fee, provided this notice is carried forward — which is what this repository does.
Commercial use and distribution is restricted to BDD and BAIR Commons members and their affiliates. Contact UC Berkeley's Office of Technology Licensing (contact details above) for commercial licensing.
This repository, and any further redistribution of it, must carry this same notice.

Citation

If you use this dataset, please cite:

@inproceedings{yu2020bdd100k,
  title={BDD100K: A Diverse Driving Dataset for Heterogeneous Multitask Learning},
  author={Yu, Fisher and Chen, Haofeng and Wang, Xin and Xian, Wenqi and Chen, Yingying and Liu, Fangchen and Madhavan, Vashisht and Darrell, Trevor},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={2633--2642},
  year={2020}
}

Acknowledgements

We sincerely thank Fisher Yu, Haofeng Chen, Xin Wang, Wenqi Xian, Yingying Chen, Fangchen Liu, Vashisht Madhavan, and Trevor Darrell, and UC Berkeley / BAIR, for creating and publicly releasing this valuable driving-scene benchmark.

Downloads last month: -