10:00 - 17:00 | Mon 29 Oct | Foyer | A1P-E
This paper aims to explore the promising possibilities offered by an RGB-D camera to address the automatic posture recognition problem. For this purpose, we designed two methods using different types of visual data provided by an RGB-D camera. The first method is based on CNN classification of 2D images, while the second uses 3D body modeling to compute and classify high level features. The experimental results demonstrated comparable performances and high precision for both methods, with a slight superiority for the CNN-based method when applied on depth images.
No information added