Multi-pose Face Recognition for Person Retrieval in Camera Networks

Abstract

In this paper, we study the use of facial appearance features for the re-identification of persons using distributed camera networks in a realistic surveillance scenario. In contrast to features commonly used for person reidentification, such as whole body appearance, facial features offer the advantage of remaining stable over much larger intervals of time. The challenge in using faces for such applications, apart from low captured face resolutions, is that their appearance across camera sightings is largely influenced by lighting and viewing pose. Here, a number of techniques to address these problems are presented and evaluated on a database of surveillance-type recordings. A system for online capture and interactive retrieval is presented that allows to search for sightings of particular persons in the video database. Evaluation results are presented on surveillance data recorded with four cameras over several days. A mean average precision of 0.60 was achieved for inter-camera retrieval using just a single track as query set, and up to 0.86 after relevance feedback by an operator.

Using latent features for short-term person re-identification with RGB-D cameras

“…Although part-based methods are very promising, holistic methods are still more robust in challenging scenarios [17]. Some other authors use other passive biometrics such as face [10], gait [46], and iris [42]. However, the low resolution of the images, occlusions, and the variety of poses make biometric-based techniques ineffective in many scenarios.…”

Who is who at different cameras: people re-identification using depth cameras

“…The use of traditional biometric modalities such as face is generally limited because it is quite common that cameras are set up to cover a relatively large area and generally there is not enough resolution to perform face recognition. One of the few approaches that uses faces is presented in [14]. In this work the cameras are set up to cover a narrow area in corridors.…”

Person re-identification in crowd

“…As an alternative, a classifier can be trained to learn the changes between cameras using labeled features. Support Vector Machines (SVM) can be employed with DCT features (Bauml et al, 2010) and SIFT (Teixeira and Corte-Real, 2009). An improvement is the Ensemble SVM, which reduces the computational cost of rankSVM for high-dimensional feature spaces besides converting the re-identification problem into a ranking problem (Prosser et al, 2010).…”

“…Similarly, Recurrent High-Structured Patches (RHSP) extract the most common blobs in the person patch (Farenzena et al, 2010); in addition to this, salient spatiotemporal edges (edgels) obtained from watershed segmentation carry information of the dominant boundary and of ratios between RGB channels (Gheissari et al, 2006). The distribution of spatial patches can be directly extracted in the frequency domain where Discrete Cosine Transform (DCT) coefficients can be used as textural features (Bauml et al, 2010). Finally, spatial patch distribution can be extracted by computing the first and the second derivatives of the person patch resulting in a covariance matrix (Wang et al, 2007;Kuo et al, 2010).…”