BDD100K: A Diverse Driving Video Database with Scalable Annotation Tooling

By Fisher Yu, Wenqi Xian, Yingying Chen, Fangchen Liu, Mike Liao, Vashisht Madhavan, Trevor Darrell. 2018

960 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Evaluation & Benchmarks

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Evaluation & Benchmarks

Citations in the corpus

A Simple Framework for Open-Vocabulary Segmentation and Detection (2023, methodology): “Following [61], we cover six commonly used segmentation datasets, including indoor scenes (ADE20K [58]), outdoor scenes (Cityscapes [6]), and driving scenes (BDD100K [52]).”
Generalized Decoding for Pixel, Image, and Language (2022, background): “[90] Fisher Yu, Wenqi Xian, Yingying Chen, Fangchen Liu, Mike Liao, Vashisht Madhavan, and Trevor Darrell.”
Rescaling Egocentric Vision: Collection, Pipeline and Challenges for EPIC-KITCHENS-100 (2020, background): “with multiple annotations and tasks, towards learning intermediate representations through downstream and multi-task learning on the same input. This has been recently achieved for autonomous driving [17,18,19,20,21,22,23,24] and scene understanding [25,26]. For example, Taskonomy [25] contains 26 t”
WoodScape: A Multi-Task, Multi-Camera Fisheye Dataset for Autonomous Driving (2019, background): “Task/Info Quantity KITTI [17] Cityscapes [10] Mapillary [39] nuScenes [6] ApolloScape [24] BDD100k [59] WoodScape Ours”
Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems (2020, background): “, 2017) and BDD100K (Yu et al., 2018) are both large video datasets containing thousands of hours of real-life driving activity.”
Benchmarking Robustness in Object Detection: Autonomous Driving when Winter is Coming (2019, methodology): “BDD100k BDD100k Yu et al. [2018] is a driving dataset consisting of 100 thousand videos of driving scenes recorded in varying conditions including weather changes and different times of the day 6 .”
MOPO: Model-based Offline Policy Optimization (2020, background): “Reinforcement learning (RL) methods, in contrast, struggle to scale to many real-world applications, e.g., autonomous driving [72] and healthcare [21], because they rely on costly online trial-and-error.”
A Survey of Autonomous Driving: Common Practices and Emerging Technologies (2019, background): “Berkeley DeepDrive [271] is a recent dataset with annotated image data.”
Stabilizing Off-Policy Q-Learning via Bootstrapping Error Reduction (2019, background): “In some domains, such as autonomous driving [38] and recommender systems [3], previously collected datasets are plentiful.”
nuScenes: A Multimodal Dataset for Autonomous Driving (2019, background): “CamVid [8], Cityscapes [19], Mapillary Vistas [33], D(2)-City [11], BDD100k [85] and Apolloscape [41] released ever growing datasets with segmentation masks.”
MSeg: A Composite Dataset for Multi-Domain Semantic Segmentation (2020, methodology): “We present MSeg, a composite dataset that uni-ﬁes semantic segmentation datasets from different domains: COCO [19], ADE20K [46], Mapillary [25], IDD [40], BDD [43], Cityscapes [7], and SUN RGB-D [36].”
A survey of deep learning techniques for autonomous driving (2019, methodology): “ data was collected. Recently, the companies Scale® and nuTonomy® started to create one of the largest and most detailed self-driving dataset on the market to date6. This includes Berkeley DeepDrive (Yu et al., 2018a), a dataset developed by researchers at Berkeley University. More relevant datasets”
Domain Randomization and Pyramid Consistency: Simulation-to-Real Generalization Without Accessing Target Domain Data (2019, methodology): “To select a model for a particular real-world dataset DR (e.g. Cityscapes), we randomly pick up 500 images from the training set of another real-world dataset DR0 (e.g. BDDS) as the validation set.”
Joint Monocular 3D Vehicle Detection and Tracking (2018, methodology): “It is laborious and expensive to annotate a large-scale 3D bounding box image dataset even in the presence of LiDAR data, although it is much easier to label 2D bounding boxes on tens of thousands of videos [47].”
ALL Snow Removed: Single Image Desnowing Algorithm Using Hierarchical Dual-tree Complex Wavelet Representation and Contradict Channel Loss (2021, neutral): “6 Semantic Segmentation- Algorithm: DeepLab [15]; Dataset: BDD [16] mIoU/mPA(%) wDH 80.”
A2D2: Audi Autonomous Driving Dataset (2020, background): “The Berkeley Deep Drive dataset (BDD-100k) [8] has a stronger emphasis on 2D bounding boxes but also contains pixel-wise segmentation annotations for 10,000 images.”
Future Event Prediction: If and When (2019, methodology): “We build on the BDD100k dataset [30], which consists of 100,000 video sequences each 40 seconds of length, accompanied with basic sensory data such as GPS, velocity or acceleration.”
Predictive Inequity in Object Detection (2019, methodology): “Finally, the BDD100K challenge (Yu et al., 2018) from which we derive our benchmark focuses on a stricter localization goal of AP 75 , corresponding to T = 0 .”
Data-free Knowledge Distillation for Object Detection (2021, methodology): “They represent the scenario where we are aware of the Dataset # required images dataset distribution performance (mAP) original teacher MS-COCO 117k original training set 0.608 distillation to student (reinitialized) ImageNet [6] 120k same-domain 0.466 VOC [9] 22k same-domain 0.443 BDD100k [45] 160k”