Classifying Tracked Moving Objects in Outdoor Urban Scenes

Object classification in far-field video sequences is a challenging problem because of low resolution imagery and projective image distortion. We approach the problem by identifying discriminative object features for classifying vehicles and pedestrians in far-field video captured by a static, uncalibrated camera. Using these features, we design a scene-invariant classification system that is trained on a small number of labelled examples from a few scenes. Simultaneously, we demonstrate that use of scene-specific features (such as image position of objects) can greatly improve classification in any given scene. To combine these ideas, we propose an algorithm for adapting a scene-invariant support vector machine object classifier to scenespecific features by retraining with the help of unlabelled data in a novel scene. Experimental results demonstrate the effectiveness of our context features and scene-transfer/adaptation algorithm for multiple urban and highway scenes.