Abstract. This paper describes a model-assisted system for reconstruction of 3D faces from a single consumer quality camera using a structure from motion approach. Typical multi-vi...
This paper presents a new algorithm for video-object segmentation, which combines motion-based segmentation, high-level object-model detection, and spatial segmentation into a sin...
Dirk Farin, Peter H. N. de With, Wolfgang Effelsbe...
We propose a novel method to automatically detect and extract the video modality of the sound sources that are present in a scene. For this purpose, we first assess the synchrony...
This work presents a real-time system for multiple objects tracking in dynamic scenes. A unique characteristic of the system is its ability to cope with longduration and complete ...
We propose a compressed domain video object segmentation method for MPEG or MPEG-like encoded videos. Computational superiority is the main advantage of the compressed domain proc...
Fatih Murat Porikli, Faisal I. Bashir, Huifang Sun