MVOR, short for Multi-View Operating Room, is the first public dataset recorded during actual clinical interventions. The dataset consists of 732 synchronized multi-view frames recorded by three RGB-D cameras. The dataset also includes visual challenges present in such environments, such as occlusion and clutter. The dataset also provides camera calibration parameters, color and depth boxes, human bounding boxes, and 2D/3D pose annotations.

Citation

@InProceedings{srivastav2018mvor, title = “MVOR: A Multi-view RGB-D Operating Room Dataset for 2D and 3D Human Pose Estimation”, author = “Srivastav, Vinkle and Issenhuth, Thibaut and Kadkhodamohammadi Abdolrahim and de Mathelin, Michel and Gangi, Afshin and Padoy, Nicolas”, conference = “MICCAI-LABELS”, year = “2018” }