Metric learning methods, for person re-identification, estimate a scaling for distances in a vector space that is optimized for picking out observations of the same individual. This paper presents a novel approach to the pedestrian re-identification problem that uses metric learning to improve the state-of-the-art performance on standard public datasets. Very high dimensional features are extracted from the source color image. A first processing stage performs unsupervised PCA dimensionality reduction, constrained to maintain the redundancy in color-space representation. A second stage further reduces the dimensionality, using a Local Fisher Discriminant Analysis defined by a training set. A regularization step is introduced to avoid singular matrices during this stage. The experiments conducted on three publicly available datasets confirm that the proposed method outperforms the state-of-the-art performance, including all other known metric learning methods. Further-more, the method is an effective way to process observations comprising multiple shots, and is non-iterative: the computation times are relatively modest. Finally, a novel statistic is derived to characterize the Match Characteristic: the normalized entropy reduction can be used to define the 'Proportion of Uncertainty Removed' (PUR). This measure is invariant to test set size and provides an intuitive indication of performance.

/pdf/local-fisher-discriminant-analysis-for-pedestrian-re-2vk7l3e6vk.pdf

Local Fisher Discriminant Analysis for Pedestrian Re-identification

Thermal cameras are passive sensors that capture the infrared radiation emitted by all objects with a temperature above absolute zero. This type of camera was originally developed as a surveillance and night vision tool for the military, but recently the price has dropped, significantly opening up a broader field of applications. Deploying this type of sensor in vision systems eliminates the illumination problems of normal greyscale and RGB cameras. This survey provides an overview of the current applications of thermal cameras. Applications include animals, agriculture, buildings, gas detection, industrial, and military applications, as well as detection, tracking, and recognition of humans. Moreover, this survey describes the nature of thermal radiation and the technology of thermal cameras.

/pdf/thermal-cameras-and-applications-a-survey-4wd1xvg4uy.pdf

Thermal cameras and applications: a survey

http://romisatriawahono.net/lecture/rm/survey/computer%20vision/Gala%20-%20Person%20Reidentification%20-%202014.pdf

Editor's Choice Article: A survey of approaches and trends in person re-identification

Person re-identification (Re-ID) is an important problem in video surveillance, aiming to match pedestrian images across camera views. Currently, most works focus on RGB-based Re-ID. However, in some applications, RGB images are not suitable, e.g. in a dark environment or at night. Infrared (IR) imaging becomes necessary in many visual systems. To that end, matching RGB images with infrared images is required, which are heterogeneous with very different visual characteristics. For person Re-ID, this is a very challenging cross-modality problem that has not been studied so far. In this work, we address the RGB-IR cross-modality Re-ID problem and contribute a new multiple modality Re-ID dataset named SYSU-MM01, including RGB and IR images of 491 identities from 6 cameras, giving in total 287,628 RGB images and 15,792 IR images. To explore the RGB-IR Re-ID problem, we evaluate existing popular cross-domain models, including three commonly used neural network structures (one-stream, two-stream and asymmetric FC layer) and analyse the relation between them. We further propose deep zero-padding for training one-stream network towards automatically evolving domain-specific nodes in the network for cross-modality matching. Our experiments show that RGB-IR cross-modality matching is very challenging but still feasible using the proposed model with deep zero-padding, giving the best performance. Our dataset is available at http:// isee.sysu.edu.cn/project/RGBIRReID.htm.

/pdf/rgb-infrared-cross-modality-person-re-identification-3yglsxf0js.pdf

RGB-Infrared Cross-Modality Person Re-identification

This paper proposes a new descriptor for person re-identification building on the recent advances of Fisher Vectors. Specifically, a simple vector of attributes consisting in the pixel coordinates, its intensity as well as the first and second-order derivatives is computed for each pixel of the image. These local descriptors are turned into Fisher Vectors before being pooled to produce a global representation of the image. The so-obtained Local Descriptors encoded by Fisher Vector (LDFV) have been validated through experiments on two person re-identification benchmarks (VIPeR and ETHZ), achieving state-of-the-art performance on both datasets.

/pdf/local-descriptors-encoded-by-fisher-vectors-for-person-re-2f71nkkrcb.pdf

Local descriptors encoded by fisher vectors for person re-identification

One of the main challenges in computer vision is the automatic detection of specific object classes in images. Recent advances of object detection performance in the visible spectrum encourage the application of these approaches to data beyond the visible spectrum. In this paper, we show the applicability of a well known, local-feature based object detector for the case of people detection in thermal data. We adapt the detector to the special conditions of infrared data and show the specifics relevant for feature based object detection. For that, we employ the SURF feature detector and descriptor that is well suited for infrared data. We evaluate the performance of our adapted object detector in the task of person detection in different real-world scenarios where people occur at multiple scales. Finally, we show how this local-feature based detector can be used to recognize specific object parts, i.e., body parts of detected people.

/pdf/feature-based-person-detection-beyond-the-visible-spectrum-4tt5yuhsg4.pdf

Feature based person detection beyond the visible spectrum

In this paper, we present an approach for person re-identification in multi-camera networks. This approach employs the Implicit Shape Model and SIFT features for person re-identification. One important property of the re-identification approach is that it is closely coupled to a person detection and tracking and uses SIFT feature models which are built during the tracking. We hold this coupling to be an important point because re-identification depends on models that are to be acquired during tracking. These models are then used to re-identify a person when it reappears in the system's field of view. Re-identification itself is performed in a 3-staged approach which allows for efficient re-identification and is perfectly suited for distributed processing where bandwidth concerns are relevant. We show that this re-identification approach — which was formerly only evaluated for single camera person re-identification can be successfully applied to the task of multi-camera re-identification. Evaluation in a challenging real-world multi-camera scenario shows that the generic approach which does not use color or other sensor specific features and thus is applicable independently of such sensor specifics — shows performance at least comparable to specialized state-of-the-art approaches.

/pdf/person-re-identification-in-multi-camera-networks-21j6n2591e.pdf

Person re-identification in multi-camera networks

A well known problem in computer vision and photogrammetry is the precise online mapping of the surrounding scenery. Due to the nature of single projective sensor configurations with inherent 7-DoF, error accumulation and scale drift is still a problem for vision based systems. This is especially relevant for difficult motion trajectories. However, it is desirable to use cheap small form factor systems, e.g., small UAVs with a single camera setup.

Monocular Camera Trajectory Optimization using LiDAR data

In this paper, we address the task of appearance basedperson reidentification in infrared image sequences. Whilecommon approaches for appearance based person reidentificationin the visible spectrum acquire color histograms ofa person, this technique is not applicable in infrared for obviousreasons. To tackle the more difficult problem of personreidentification in infrared, we introduce an approachthat relies on local image features only and thus is completelyindependent of sensor specific features which mightbe available only in the visible spectrum. Our approachfits into an Implicit Shape Model (ISM) based person detectionand tracking strategy described in previous work.Local features collected during tracking are employed forperson reidentification while the generalizing appearancecodebook used for person detection serves as structuringelement to generate person signatures. By this, we gain anintegrated approach that allows for fast online model generation,a compact representation, and fast model matching.Since the model allows for a joined representation ofappearance and spatial information, no complex representationmodels like graph structures are needed. We evaluateour person reidentification approach on a subset of the CASIAinfrared dataset.

/pdf/local-feature-based-person-reidentification-in-infrared-57k8486dqi.pdf

Local Feature Based Person Reidentification in Infrared Image Sequences

In this paper, we approach the task of appearance based person re-identification for scenarios where no biometric features can be used. For that, we build on a person re-identification approach that uses the Implicit Shape Model (ISM) and SIFT features for re-identification. This approach builds identity models of persons during tracking and employs these models for re-identification. We apply this re-identification, which was until now only evaluated in the infrared spectrum, to data acquired in the visible spectrum. Furthermore we evaluate view independence of the re-identification approach and introduce methods that extend view invariance. Specifically, we (i) propose a method for online view-determination of a tracked person, (ii) use the online view-determination to generate view specific identity models of persons which increase model distinctiveness in re-identification, and (iii) introduce a method to convert identity models between views to increase view independence.

/pdf/view-invariant-person-re-identification-with-an-implicit-4i5d6wipn0.pdf

Kai Jungling

Papers

Feature based person detection beyond the visible spectrum

Person re-identification in multi-camera networks

Monocular Camera Trajectory Optimization using LiDAR data

Local Feature Based Person Reidentification in Infrared Image Sequences

View-invariant person re-identification with an Implicit Shape Model