Noldus plays an active role in research and collaboration projects around the world. Often these result in concepts or prototypes of new tools for behavioral research that will benefit the research community in future. For developing and validating new algorithms for automatic behavior recognition, often large sets of videos, calibration data, and observational data are collected during these projects.
Because Noldus wants to contribute to the scientific community, we also want to make these sets available to other scientists around the world. At this moment the following sets are available:
iEatSet: Eating Behavior Dataset
The iEatSet includes 75 instances captured from 15 participants consuming 5 types of meals. Each session is recorded by a frontally placed Kinect for Windows v2 sensor, a sideview and a topview camera and 4 on-body motion sensors.
The dataset includes eating gestures in seven The Observer XT projects (version 12) and rough data (accelerometer, magnetometer, gyrometer, videos, annotations and pointcloud data from the Kinect sensor).
RatSI: Rat Social Interaction Dataset
The Rat Social Interaction (RatSI) dataset comprises 9 fully annotated videos of two rats interacting socially in a 90x90cm Noldus PhenoTyper™ 9000 cage without bedding and accessories. The recordings are made from top-view perspective and are about 15 minutes long. We provide both tracking data and frame-by-frame annotations that were scored by an expert in rat behavior.