,, Dhiraj and Sumeet , M and Somsukla, S and Sanja, M
(2018)
Activity Recognition for Indoor Fall Detection in 360- degree videos using Deep Learning techniques.
In: Third International Conference on Comfuter Vision & Image Procsing (CVIP) ,2018, September 29 - October 1, 2018, IIITDM , Jabalpur .
(Submitted)
Abstract
Abstract. l luman activity rccognition(I-IAR) targets the methodologies to rec ognize the different actions from a sequence of observations. Vision-based ac tivity recognition is among the most popular unobtrusive technique for activity recogn ition. Caring for elderly, living alone from remotely is one of the biggest challenges of modem human society and is an area of active research. Us age of smart homes with an increasing number of cameras in our daily environ ment provides the platform to use that technology for activity recognition also. The usage of omnidirectional cameras for fall detection activity minimizes the requirement of multiple cameras for fall detection in the living scenario. Conse quen tly , we propose a vision-based solution using convolutional neural net works and long short-term memory networks using 360-degree videos for hu man fall detection. We have constructed an omnidirectional video dataset by re cording a set of activities performed by different people as no such 360-degree video dataset is available for human activity recognition in the public domain. Our results prove the suitability of 3DCNN and LSTM techniques for fall detection activity for even omnidirectional videos.
Actions (login required)
 |
View Item |
Download Statistics
Download Statistics