Gesture recognition from video sequences

Abstract : In this thesis, we aim to recognize gestures (e.g. hand raising) and more generally short actions (e.g. fall, bending) accomplished by an individual. Many techniques have already been proposed for gesture recognition in specific environment (e.g. laboratory) using the cooperation of several sensors (e.g. camera network, individual equipped with markers). Despite these strong hypotheses, gesture recognition is still brittle and often depends on the position of the individual relatively to the cameras. We propose to reduce these hypotheses in order to conceive general algorithm enabling the recognition of the gesture of an individual involving in an unconstrained environment and observed through limited number of cameras. The goal is to estimate the likelihood of gesture recognition in function of the observation conditions. Our method consists of classifying a set of gestures by learning motion descriptors. These motion descriptors are local signatures of the motion of corner points which are associated with their local textural description. We demonstrate the effectiveness of our motion descriptors by recognizing the actions of the public databases KTH and IXMAS.
Complete list of metadatas

Cited literature [95 references]  Display  Hide  Download
Contributor : Francois Bremond <>
Submitted on : Wednesday, November 4, 2009 - 6:23:39 PM
Last modification on : Saturday, January 27, 2018 - 1:30:46 AM
Long-term archiving on : Saturday, November 26, 2016 - 1:20:00 PM


  • HAL Id : tel-00428690, version 2



Mohamed Kaâniche. Gesture recognition from video sequences. Signal and Image processing. Université Nice Sophia Antipolis, 2009. English. ⟨tel-00428690v2⟩



Record views


Files downloads