Probabilistic Models and Inference for Multi-View People Detection in Overlapping Depth Images

Mehr zum Buch

Focusing on advanced techniques for indoor people detection, this work explores the integration of multi-view information from depth sensors to enhance detection accuracy. It addresses the challenges of overlapping depth images by reformulating the detection task as an inverse problem. The authors introduce a generative probabilistic framework that utilizes both temporal context and multi-view evidence, aiming to improve performance in wide-area scenarios. This approach highlights the significance of leveraging redundant and complementary data for effective detection.