Pedestrian Detection In Crowded Scenes Seminar Mustererkennung

This distribution addresses the problem of detecting pedestrians in crowded real-world scenes with severe overlaps. The basic premise is that this problem is too difficult for any type of model or feature alone. The first algorithm that integrates evidence in multiple iterations and from different sources proposed by Leibe et al. [2005] is presented. The core part of this method is the combination of local and global cues via a probabilistic top-down segmentation. Altogether, this approach allows to examine and compare object hypotheses with high precision down to the pixel level. Qualitative and quantitative results on a large data set confirm that their method is able to reliably detect pedestrians in crowded scenes. The second work presented, is the person detection system proposed by Wu and Nevatia [2005]. They learn multiple part detectors for full body, head-shoulder, torso and legs and presented a novel edge-based feature called edgelet. The authors show that edgelet features perform better than Haar-wavelet features in their boosting framework. Part hypotheses are aggregated in a probabilistic formulation with a Gaussian assumption. They quantitatively evaluate on a surveillance task, but show also impressive results on other types of images.