A Survey on RGB-D Datasets




In this page, we will list all datasets presents in our Paper “A Survey on RGB-D Datasets”. To cite our work, just add:

@misc{lopes2022survey,
      title={A Survey on RGB-D Datasets}, 
      author={Alexandre Lopes and Roberto Souza and Helio Pedrini},
      year={2022},
      eprint={2201.05761},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}

If you have any comments or want to add your work to the list, just contact us at:

  • alexandre.lopes at ic.unicamp.br or
  • alexrlops at gmail.com

Also check our Github Awesome Page:


Updates

25/05/22 – Inclusion of 28 new datasets of Saliency Object Detection and Action Recognition. We reviewed extra 606 papers to get to this list extension.



Datasets’ List Presented in the Paper

FILTER
Dataset NameAcquisition typeSensorApplicationAreaData ModalitiesExtra DataImages/ScenesYear
--SyntheticSOR and SOEAerialColor, DepthNormal Maps, Edges, Semantic Labels15 scenes (144000 images)2020
SBM-RGBD datasetStructured LightKinect v1SOEIndoorColor, DepthSaliency Mask33 sequences (~15000 frames)2017
LFSD-Lytro light field (MVS)SOEIn-The-WildColor, DepthSaliency Mask100 images2015
An In Depth View of SaliencyStructured LightKinect v1SOEIndoorColor, DepthSaliency Mask80 images2013
DUTLF-Depth-Lytro light field (MVS)SOEIn-The-WildColor, DepthSaliency Mask1200 images2019
ReDWeb-S-MVSSOEIn-The-WildColor, DepthSaliency Mask3179 images2020
COTS-Intel Realsense D435 (MVS)SOEIsolated Objects / Focussed on ObjectsColor, DepthSaliency Mask120 images2021
NTU RGB+DTOFKinect v2Human ActivitiesFull BodyColor, Depth, IRPerson Pose (Skeleton)56880 sequences2016
NTU RGB+D 120TOFKinect v2Human ActivitiesFull BodyColor, Depth, IRPerson Pose (Skeleton)111480 sequences2019
Mivia ActionStructured LightKinect v1Human ActivitiesFull BodyColor, Depth-28 sequences2013
Chalearn LAP IsoGDStructured LightKinect v1Human ActivitiesFull BodyColor, Depth-47933 sequences2016
SYSU 3D HOIStructured LightKinect v1Human ActivitiesFull BodyColor, DepthPerson Pose (Skeleton)480 sequences2015
G3DStructured LightKinect v1Human ActivitiesFull BodyColor, DepthPerson Pose (Skeleton), Semantic Segmentation7 sequences (multiple actions per sequence)2016
IAS-Lab RGBD-IDStructured LightKinect v1Human ActivitiesFull BodyColor, DepthPerson Pose (Skeleton), Semantic Segmentation33 sequences2013
Online RGBD Action Dataset (ORGBD)Structured LightKinect v1Human ActivitiesFull BodyColor, DepthPerson Pose (Skeleton)48 sequences2014
MADStructured LightKinect v1Human ActivitiesFull BodyColor, DepthPerson Pose (Skeleton)40 sequences2014
Hand GestureStructured LightKinect v1Gestures (Head and Hand)Partial Body w/o SceneColor, Depth-1400 sequences2014
Creative Senz3DStructured LightCreative Senz3DGestures (Head and Hand)Partial Body w/o SceneColor, Depth-1320 sequences2015
PKU-MMDTOFKinect v2Human ActivitiesFull BodyColor, DepthPerson Pose (Skeleton)3076 sequences2017
Florence 3D ActionsStructured LightKinect v1Human ActivitiesFull BodyColor, DepthPerson Pose (Skeleton)215 sequences2013