LEARNING 3D OBJECT-CENTRED APPEARANCE MODELS FOR TRACKING
暂无分享,去创建一个
This paper presents a hypothesis verification strategy for 3D object recognition. This methodology integrates 3D object-centered and 2D appearance-based representations in computer vision which leads to improved hypothesis verification. The approach is demonstrated on real-world image sequences from traffic surveillance and compared to edge-based iconic evaluation techniques.