Citation Context

...oncerned with establishing the best absolute performance given various state-of-the-art pedestrian classifiers. We refer the reader to recently proposed systems and benchmark studies, e.g., [4], [7], =-=[10]-=-, [20], [23], [48], [50], [52], [57], [60]. Rather, our aim is to demonstrate the relative performance gain resulting from the proposed multilevel approach, exemplified using state-of-the-art feature ...

Citation Context

...8, 4, 10], and detector transfer in new scenes [42, 27]. We refer the detailed surveys on pedestrian detection to [21, 14]. Resolution related problems have attracted attention in recent evaluations. =-=[16]-=- found that the pedestrian detection performance depends on the resolution of training samples. [14] pointed that the pedestrian detection performance drops with decreasing resolution. [23] observed s...

Citation Context

... comic-strips. 2 RELATED WORK Many image databases have been built to train detection and recognition algorithms. Representative databases include Yale face database [16], Daimler pedestrian database =-=[17]-=-, INRIA database for human detection [18] and HumanEva [19] for pose estimation. Typically, the image quality of these databases is limited (e.g., grayscale images of size between 200 to 300 pixels). ...

Citation Context

...s include the Middlebury data set for disparity/depth generation [16], Pascal Visual Object Classes Challenge for classification of various objects [11], and the Daimler Occluded Pedestrian benchmark =-=[10]-=-, which all focus on reocognition accuracy and present the core of their results using precision versus recall curves. While accuracy is important for the general problem set, the mobile space introdu...

Citation Context

...ound, or might be partially occluded. Most state-of-the-art systems derive feature sets from intensity images, i.e. gray-scale (or colour) images, and apply learning-based approaches to detect people =-=[1, 3, 9, 22, 23]-=-. Besides image intensity, depth information can provide additional cues for pedestrian recognition. Up to now, the use of depth information has been limited to recovering high-level scene geometry [5...

Citation Context

...ariant of the well-known HOG-based pedestrian detector, e.g. [4], in both monocular and stereo vision set-ups. We assume our results to generalize to other established pedestrian detectors, e.g. [5], =-=[7]-=-, [10], [13], [16], [17]. A second contribution involves a new large real-world stereo dataset for pedestrian detection which is used in our experiments. We make this dataset publicly available for no...

Citation Context

...ork, over the last decade, a considerable number of vision-based pedestrian detection systems have been proposed. Several remarkable surveys have been presented [1-4], some of them recently published =-=[5,6]-=-. Most of the work concerning human motion summarized in [1-5] focuses on pedestrian protection applications in the intelligent vehicle domain, covering both passive and active safety techniques. An o...

Citation Context

...ure 2: Example detections of a top performing method (SquaresChnFtrs). 2 Datasets Multiple public pedestrian datasets have been collected over the years; INRIA [1], ETH [2], TUD-Brussels [3], Daimler =-=[4]-=- (Daimler stereo [5]), Caltech-USA [6], and KITTI [7] are the most commonly used ones. They all have different characteristics, weaknesses, and strengths. INRIA is amongst the oldest and as such has c...

Citation Context

...interested in both the ROC performance (trade-off correct vs. false detections) and in the processing cost. 2 Related Work Many interesting approaches for pedestrian detection have been proposed. See =-=[4]-=- for a recent survey and a novel publicly available benchmark set. Most work has proceeded with a learning-based approach by-passing a pose recovery step and describing human appearance directly in te...