└── README.md /README.md: -------------------------------------------------------------------------------- 1 | ## Datasets for nighttime and low-light conditions. 2 | 3 | ### Object Detection 4 | - [ExDark (Exclusively Dark Image Dataset)](https://github.com/cs-chan/Exclusively-Dark-Image-Dataset) 5 | -- Labelled 6 | -- Multiple objects are annotated. 7 | - [BDD100K: Night](https://bdd-data.berkeley.edu/) 8 | -- Labelled (Small Data) 9 | -- Cars and other vehicles are annotated. 10 | 11 | ### Face Detection 12 | - [DarkFace Dataset](https://flyywh.github.io/CVPRW2019LowLight/) 13 | -- Labelled 14 | -- 5400/600 images for train/val 15 | -- Face detection in extremely low-light 16 | 17 | ### People/Pedestrian Detection 18 | - [NightOwls Dataset](https://www.nightowls-dataset.org/) 19 | -- Labelled 20 | -- Pedestrian detection 21 | - [KAIST Multi-Spectral Dataset](https://soonminhwang.github.io/rgbt-ped-detection/data/) 22 | -- Labelled 23 | -- Pedestrian detection 24 | 25 | 26 | ### Pose Estimation 27 | - [ExLPose Dataset](http://cg.postech.ac.kr/research/ExLPose/) 28 | -- Labelled 29 | -- Human 2D Pose estimation in extremely low-light 30 | 31 | ### Stereo Vision 32 | - [The Oxford RobotCar](https://robotcar-dataset.robots.ox.ac.uk/) 33 | -- Unlabelled 34 | -- Provides raw LIDAR based sparse depth maps. 35 | - [DENSE: Gated2Depth](https://www.uni-ulm.de/en/in/driveu/projects/dense-datasets) 36 | -- Unlabelled 37 | -- Provides raw LIDAR based sparse depth maps 38 | - [DENSE: PixelAccurateDepthBenchmark](https://www.uni-ulm.de/en/in/driveu/projects/dense-datasets) 39 | -- Labelled (Small Data) 40 | -- 4 scenes with 10 images each. Each image comes with a pixel-accurate dense depth map. 41 | 42 | ### Semantic Segmentation 43 | - [BDD100K: Night](https://bdd-data.berkeley.edu/) 44 | -- Labelled (Small Data) 45 | -- Semantic classes compatible with the [CityScapes](https://www.cityscapes-dataset.com/) dataset 46 | - [Dark Zurich](https://www.trace.ethz.ch/publications/2019/GCMA_UIoU/) 47 | -- Unlabelled 48 | -- Provides a small set of 50 labelled images for validation/test purposes. 49 | -- Semantic classes compatible with the [CityScapes](https://www.cityscapes-dataset.com/) dataset 50 | - [NightCity](https://arxiv.org/pdf/2003.06883.pdf) 51 | -- Yet to be published 52 | --------------------------------------------------------------------------------