Close-Range Human Detection and Tracking for Head-Mounted Cameras

Abstract

In this paper we consider the problem of multi-person detection from the perspective of a head mounted stereo camera. As pedestrians close to the camera cannot be detected by classical full-body detectors due to strong occlusion, we propose a stereo depth-template based detection approach for close-range pedestrians. We perform a sliding window procedure, where we measure the similarity between a learned depth template and the depth image. To reduce the search space of the detector we slide the detector only over few selected regions of interest that are generated based on depth information. The region-of-interest selection allows us to further constrain the number of scales to be evaluated, significantly reducing the computational cost. We present experiments on stereo sequences recorded from a head-mounted camera setup in crowded shopping street scenarios and show that our proposed approach achieves superior performance on this very challenging data.

True scaled 6 DoF egocentric localisation with monocular wearable systems

“…Most works on wearable vision systems have developed towards recognition of human activities, where many approaches take advantage from a nice feature of chest-wearable cameras: the prior knowledge of the action or manipulation taking place at the centre of the image. Recognition problems where wearable systems have been applied are segmentation of handled objects [18,19], recognition of activities and objects in the workspace [20], novelty detection in a daily routine [21], clustering of sport activities from video sequences [22], human detection [23] or analysis of human movements to infer social interactions [24].…”

An ultra-fast human detection method for color-depth camera

“…Head-shoulder template matching [19,27,23] is widely used for detecting human beings, but it often fails in handling partially observed subjects. We propose a novel knowledge based human locator using shape information of human body which can deal with partial occlusion and depth data loss to filter the results produced by the previous stage.…”

“…A top-down view setup [18] is a good choice for decreasing overlapping and facilitating segmentation, but the visual field is limited and details of human subject cannot be well observed. A smart methodology taking advantage of top-down view while maintaining sufficient visual details is adopted by [19] and our previous work [20]. These methods take a front (or oblique) view, and project the 3D point cloud of the scene onto the ground plane to obtain a virtual plan view.…”

Detecting and tracking people in real time with RGB-D camera

“…The method proposed by Mitzel and Leibe (2012) uses a learned upper-body depth template and can work well in crowded scenarios, but it requires multi-scale downsampling (sampling rate will restrict the accuracy and operation efficiency). The system of Bahadori et al (2007) exploits the correspondence between image blobs and world blobs, which can alleviate occlusions in original images but still faces difficulties in blob segmentation when people are highly crowded.…”