Skip to Main content Skip to Navigation
Theses

Reconnaissance d'événements et d'actions à partir de la profondeur thermique 3D

Adnan Al Alwani 1
1 Equipe Image - Laboratoire GREYC - UMR6072
GREYC - Groupe de Recherche en Informatique, Image et Instrumentation de Caen
Abstract : In this thesis, we deal with the problem of detection of flows events from thermal cameras and Kinect depth cameras. The first part concerns the development of a thermal event recognition platform. We are particularly interested in the detection of pain in preterm babies and we have introduced a new database (Pretherm) carried out within the framework of the ANR project of the same name, with videos of premature babies from CHU hospital of Caen. To characterize the pattern of interest, we exploited the non-redundant (or residual) local binary patterns. As well as the topological persistence of the monodimensional signal. Validation was performed using Extreme Learning Machines and Support Vector Machines. In the second part, we studied the problem of recognition with Kinect. We were particularly interested in spatio-temporal descriptors and studied the classification methods dedicated to action recognition. Initially, we focused on the analysis of the angles of the joints and their trajectories using the HMMs. Then we proposed a compact representation using spherical harmonics to learn and recognize the poses. These proposed techniques have been validated on a wide variety of RGBD corpus.
Complete list of metadata

Cited literature [31 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/tel-01418369
Contributor : Greyc Référent Connect in order to contact the contributor
Submitted on : Friday, December 16, 2016 - 4:24:09 PM
Last modification on : Tuesday, October 19, 2021 - 11:34:59 PM
Long-term archiving on: : Tuesday, March 21, 2017 - 11:45:14 AM

Identifiers

  • HAL Id : tel-01418369, version 1

Citation

Adnan Al Alwani. Reconnaissance d'événements et d'actions à partir de la profondeur thermique 3D. Vision par ordinateur et reconnaissance de formes [cs.CV]. Université de Caen Normandie, 2016. Français. ⟨tel-01418369⟩

Share

Metrics

Record views

232

Files downloads

294