This dataset is composed of RGB-D video and skeleton data of 8 social activities.
Every video contains a single activity and those are performed in repetition for 40 secs.
The list of results for this dataset can be found on this result page.
This dataset is composed of long RGB-D videos and skeleton data.
The videos contain several social and individual activities with no restriction on which one has to be performed.
The objective of this dataset is to evaluate a system for the temporal segmentation of the social activities.
The labels provided concern only the time intervals of social or the individual activities regardless of which activities are actually occurring.
UoL 3D Continuous Social Activity Dataset
Link will be provided after acceptance of the paper.
The dataset is composed of 20 long RGB-D videos. Each video provides RGB-D images and tracked skeleton of different social and individual activities performed by two people.
The labels provided concern only the intervals of the social activities.
Indeed, the dataset is built with the intention of evaluate systems for segmentation and classification of social activities.