Estimation of Locations of Densely Distributed Subjects Using NMF with Nonpixel Information

We propose a method that uses a large number of digital photographs to produce highly accurate estimates of the locations of subjects that have attracted a crowd’s attention. Recently, a very active area of research has been to use humans as sensors in realworld observations that require a large amount of data. Some of these studies have attempted to produce real-time estimates of the subjects that are attracting a crowd’s attention by quickly collecting a large number of photographs. These studies are based on the assumption that, when photographers encounter interesting events, they take pictures. Some of the proposed methods realize high availability by using only photographing information, which includes information about location and azimuth of the camera and it is automatically embedded into photograph. Since this data is very small compared to that of the pixel information, the load on the communication infrastructure is reduced. However, there are problems with the accuracy when there are many attractive subjects in a small region, and they cannot be found with traditional methods that use a sequential search strategy. The proposed method overcomes this problem by applying nonnegative matrix factorization (NMF) to the estimated location of each subject. We verified the effectiveness of this by computational experiments and an experiment under a realistic environment.

Dynamic Cell Structures for the Evaluation of Keypoints in Facial Images

In this contribution Dynamic Cell Structures (DCS network) are applied to classify local image structures at particular facial landmarks. The facial landmarks such as the corners of the eyes or intersections of the iris with the eyelid are computed in advance by a combined model and data driven sequential search strategy. To reduce the detection error after the processing of the sequential search strategy, the computed image positions are verified applying a DCS network. The DCS network is trained by supervised learning with feature vectors which encode spatially arranged edge and structural information at the keypoint position considered. The model driven localization as well as the data driven verification are based on steerable filters, which build a representation comparable with one provided by a receptive field in the human visual system. We apply a DCS based classifier because of its ability to grasp the topological structure of complex input spaces and because it has proved successful in a number of other classification tasks. In our experiments the average error resulting from false positive classifications is less than 1%.