Action Detection from Egocentric Videos in Daily Living Scenarios

Authors: Georgios Kapidis, Elsbeth van Dam, Ronald Poppe, Lucas Noldus, Remco Veltkamp

Abstract

We are researching the use of egocentric vision in the area of Human Action Recognition. Inspired from recent advances in activity recognition from video using deep learning, we investigate the detection performance of Long Short-Term Memory networks on an elementary set of Activities of Daily Living, based on the detected objects in the scene.

In proceedings of: Measuring Behavior 2018