Human model for people detection in dynamic scenes

Abstract

http://www2.computer.org/portal/web/csdl/doi/10.1109/ICPR.2006.638 http://portal.acm.org/citation.cfm?id=1172043International audienceThe problem of multiple people detection in monocular video streams is addressed. The proposed method involves a human model based on skin color and foreground information. Robustness to local motion of background and global color changes is achieved by modeling images as fields of color distributions, and robustly estimating temporal background global variations. The estimation of the human model parameters is done via Monte Carlo simulations to deal with the multimodal nature of the posterior distribution, introduced by the presence of multiple people and cluttered scene. Promising results are presented for transportation vehicles sequences

Robust People Detection by Fusion of Evidence from Multiple Methods

“…These regions are then matched to the nodes of a tree that represents the constituent parts of a person, as well as their associated topological relationships and probability distributions. Another approach described in [5] considers a model of a person constituted by three rectangles that denote the locations of the torso, head and face. The typical percentages of foreground pixels inside the three rectangles and of skin pixels in the face rectangle are precomputed based on training images.…”

Post-processing approaches for improving people detection performance

“…There are two main conventional object detection approaches: one based on some kind of segmentation of the scene in foreground (objects) and background [1]- [13] and one based on an exhaustive scanning approach [14]- [42]. There are also some approaches that try to combine both approaches together [43], [44].…”

“…We classified the appearance models according to simplified human models or complex models. There are simple person models that define the person as a region or shape, i.e., holistic models [2]- [10], [14], [15], [17]- [20], [23]- [38], [42] and more complex models that define the person as combination of multiple regions or shapes, i.e., part-based models [11]- [13], [21], [22], [25], [39]- [41], [43], [44]. Although the vast majority of approaches are mainly based on appearance information, there are some approaches that combine appearance and motion information in order to improve the detection results.…”

Smart Objects Identification System for Robotic Surveillance

“…Reference [17] proposed a method to classify human model based on skin color and foreground information which is foreground pixels and skin color probability maps. However, the skin color is difficult to be extracted when the objects are far from the camera and it will give the false results when the colors of skin and other objects are similar.…”