FUKinect-Fall dataset was created using Kinect V1. The dataset includes walking, bending, sitting, squatting, lying and falling actions performed by 21 subjects between 19-72 years of age. A total of 1008 depth videos and 3D coordinates (x, y, z) of 20 joints were recorded in total (6 actions × 8 repeat × 21 subjects). Each video duration is recorded as approximately 4-5 seconds, 320 × 240 resolution and 30 frames per second depending on the action feature.
Dataset depth video size is about 7.5GB. Therefore, the dataset is not loaded. Dataset can be downloaded from the link below
Download link
https://www.dropbox.com/s/wl1o4x2xobj1s8w/FUKinect_Fall.rar?dl=0
Please Citeation
Aslan M., Akbulut Y., Sengor A., CevdetInce M. "Skeleton based efficient fall detection", J. Faculty Eng. Architecture Gazi Univ., 32 (4) (2017), pp. 1025-1034. (DOI: 10.17341/gazimmfd.369347),