Mutual On-Line Learning for Detection and Tracking in High-Resolution Images

Abstract

This paper addresses object detection and tracking in high-resolution omnidirectional images. The foreseen application is a visual subsystem of a rescue robot equipped with an omnidirectional camera, which demands real-time efficiency and robustness against changing viewpoint. Object detectors typically do not guarantee specific frame rate. The detection time may vastly depend on a scene complexity and image resolution. The adapted tracker can often help to overcome the situation, where the appearance of the object is far from the training set. On the other hand, once a tracker is lost, it almost never finds the object again. We propose a combined solution where a very efficient tracker (based on sequential linear predictors) incrementally accommodates varying appearance and speeds up the whole process. Next we propose to incrementally update the detector with examples collected by the tracker. We experimentally show that the performance of the combined algorithm, measured by a ratio between false positives and false negatives, outperforms both individual algorithms. The tracker allows to run the expensive detector only sparsely enabling the combined solution to run in real-time on 12 MPx images from a high resolution omnidirectional camera (Ladybug3). © Springer-Verlag Berlin Heidelberg 2013.