Understanding transit scenes: a survey on human behavior-recognition algorithms
IEEE Transactions on Intelligent Transportation Systems
WD'09 Proceedings of the 2nd IFIP conference on Wireless days
Networked heterogeneous camera system for high resolution face images
ISVC'06 Proceedings of the Second international conference on Advances in Visual Computing - Volume Part II
Real-Time Query Processing on Live Videos in Networks of Distributed Cameras
International Journal of Interdisciplinary Telecommunications and Networking
Hi-index | 0.00 |
This paper presents the groundwork for a distributed network of collaborating, intelligent surveillance cameras, implemented with low-cost embedded microprocessor camera modules. Each camera trains a person detection classifier using the Winnow algorithm for unsupervised, online learning. Training examples are automatically extracted and labelled, and the classifier is then used to locate person instances. To improve detection performance, multiple cameras with overlapping fields of view collaborate to confirm results. We present a novel, unsupervised calibration technique that allows each camera module to represent its spatial relationship with the rest. During runtime, cameras apply the learned spatial correlations to confirm each other's detections. This technique implicitly handles non-overlapping regions that cannot be confirmed. Its computational efficiency is well-suited to real-time processing on our hardware.