Human detection in RGB-D images is an important yet very challenging task in computer vision. In this paper, we propose a novel human detection approach in RGB-D images, which integrates ROI (region-of-interest) generation, depth-size relationship estimation and a human detector. Our approach has the following advantages: 1) ROI generation and depth-size relationship estimation take full advantage of color and depth information to fast reject about 70% negative samples while maintaining a high recall rate; 2) the cascade-structured human detector can seamlessly concatenate features extracted from both color and depth images; and 3) our method can detect human at a speed of more than 30 fps on 640 χ 480 images on a single laptop CPU without any GPU acceleration. Experiments on challenging public datasets demonstrate the effectiveness of our method.

Massive content delivery in metropolitan networks has recently gained much attention with the successful deployment of commercial systems and an increasing user popularity. With an enormous volume of content available in the network, as well as the growing size of content owing to the popularity of high-definition video, the exploration of capacity in the caching network becomes a critical issue in providing guaranteed service. Yet, collaboration strategies among cache servers in emerging scenarios, such as IPTV services, are still not well understood so far. In this paper, we propose an efficient collaborative caching mechanism based on the topology derived from a real-world IPTV system, with a particular focus on exploring the capacity of the existing system infrastructure. We observe that collaboration among servers is largely affected by the topology characteristics and heterogeneous capacities of the network. Meanwhile, dynamic request routing within the caching network is strongly coupled with content placement decisions when designing the mechanism. Our proposed mechanism is implemented in a distributed manner, and is amenable to practical deployment. Our simulation results demonstrate the effectiveness of our proposed mechanism, as compared to conventional cache cooperation with static routing schemes.