BDD100K: A Diverse Driving Video Database with Scalable Annotation Tooling

https://arxiv.org/abs/1805.04687
http://bdd-data.berkeley.edu/

Videos 100K video clips Size:1.8TB md5: 253d9a2f9d89d2b09d8d93f397aecdd7

Info The GPS/IMU information recorded along with the videos Size: 3.9GB md5: 043811ff34b2fca6d50f37d263a65c93

Images It has two subfolders. 1) 100K labeled key frame images extracted from the videos at 10th second 2) 10K key frames for full-frame semantic segmentation. Size: 6.5GB md5: b538e3731a132e28dae37f18c442c51e

Labels Annotations of road objects, lanes, and drivable areas in JSON format. Details at Github repo. Size: 147MB md5: 035bd9e56c0c2dbdef2165e0bf1cf6a5

Drivable Maps Segmentation maps of Drivable areas. Size: 661MB md5: 1bd70019468a81572c70d374751eb9e2

Segmentation Full-frame semantic segmentation maps. The corresponding images are in the same folder. Size: 1.2GB md5: 0baeaf8bed8f1a7feb1e8755bcce7169