Srikrishna Karanam

Person Re-Identification with Discriminatively Trained Viewpoint Invariant Dictionaries

This paper introduces a new approach to address the person re-identification problem in cameras with non-overlapping fields of view. Unlike previous approaches that learn Mahalanobis-like distance metrics in some transformed feature space, we propose to learn a dictionary that is capable of discriminatively and sparsely encoding features representing different people. Our approach directly addresses two key challenges in person re-identification: viewpoint variations and discriminability. First, to tackle viewpoint and associated appearance changes, we learn a single dictionary to represent both gallery and probe images in the training phase. We then discriminatively train the dictionary by enforcing explicit constraints on the associated sparse representations of the feature vectors. In the testing phase, we re-identify a probe image by simply determining the gallery image that has the closest sparse representation to that of the probe image in the Euclidean sense. Extensive performance evaluations on three publicly available multi-shot re-identification datasets demonstrate the advantages of our algorithm over several state-of-the-art dictionary learning, temporal sequence matching, and spatial appearance and metric learning based techniques.

Sparse re-id: Block sparsity for person re-identification

This paper presents a novel approach to solve the problem of person re-identification in non-overlapping camera views. We hypothesize that the feature vector of a probe image approximately lies in the linear span of the corresponding gallery feature vectors in a learned embedding space. We then formulate the re-identification problem as a block sparse recovery problem and solve the associated optimization problem using the alternating directions framework. We evaluate our approach on the publicly available PRID 2011 and iLIDS-VID multi-shot re-identification datasets and demonstrate superior performance in comparison with the current state of the art.

Multi-Shot Human Re-Identification Using Adaptive Fisher Discriminant Analysis.

While much research in human re-identification has focused on the single-shot case, in real-world applications we are likely to have an image sequence from both the person to be matched and each candidate in the gallery, extracted from automated video tracking. It is desirable to take advantage of the multiple visual aspects (states) of each subject observed during training and testing. However, since each subject may spend different amounts of time in each state, equally weighting all the images in a sequence is likely to produce suboptimal performance. To address this problem, we introduce an algorithm to hierarchically cluster image sequences and use the representative data samples to learn a feature subspace maximizing the Fisher criterion. The clustering and subspace learning processes are applied iteratively to obtain diversity-preserving discriminative features. A metric learning step is then applied to bridge the appearance difference between two cameras. The proposed method is evaluated on three multi-shot re-id datasets and the results outperform state-of-the-art methods.

From the Lab to the Real World: Re-identification in an Airport Camera Network

Over the past ten years, human re-identification has received increased attention from the computer vision research community. However, for the most part, these research papers are divorced from the context of how such algorithms would be used in a real-world system. This paper describes the unique opportunity our group of academic researchers had to design and deploy a human re-identification system in a demanding real-world environment: a busy airport. The system had to be designed from the ground up, including robust modules for real-time human detection and tracking, a distributed, low-latency software architecture, and a front-end user interface designed for a specific scenario. None of these issues are typically addressed in re-identification research papers, but all are critical to an effective system that end users would actually be willing to adopt. We detail the challenges of the real-world airport environment, the computer vision algorithms underlying our human detection and re-identification algorithms, our robust software architecture, and the ground-truthing system required to provide the training and validation data for the algorithms. Our initial results show that despite the challenges and constraints of the airport environment, the proposed system achieves very good performance while operating in real time.

Real-World Re-Identification in an Airport Camera Network

Human re-identification across non-overlapping fields of view is one of the fundamental problems in video surveillance. While most reported research for this problem is focused on improving the matching rate between pairs of cropped rectangles around humans, the situation is quite different when it comes to creating a re-identification algorithm that operates robustly in the real world. In this paper, we describe an end-to-end system solution of the re-identification problem installed in an airport environment, with a focus on the challenges brought by the real world scenario. We discuss the high-level system design of the video surveillance application, and the issues we encountered during our development and testing. We also describe the algorithm framework for our human re-identification software, and discuss considerations of speed and matching performance. Finally, we report the results of an experiment conducted to illustrate the output of the developed software as well as its feasibility for the airport surveillance task.

DukeMTMC4ReID: A Large-Scale Multi-camera Person Re-identification Dataset

In the past decade, research in person re-identification (re-id) has exploded due to its broad use in security and surveillance applications. Issues such as inter-camera viewpoint, illumination and pose variations make it an extremely difficult problem. Consequently, many algorithms have been proposed to tackle these issues. To validate the efficacy of re-id algorithms, numerous benchmarking datasets have been constructed. While early datasets contained relatively few identities and images, several large-scale datasets have recently been proposed, motivated by data-driven machine learning. In this paper, we introduce a new large-scale real-world re-id dataset, DukeMTMC4ReID, using 8 disjoint surveillance camera views covering parts of the Duke University campus. The dataset was created from the recently proposed fully annotated multi-target multi-camera tracking dataset DukeMTMC[36]. A benchmark summarizing extensive experiments with many combinations of existing re-id algorithms on this dataset is also provided for an up-to-date performance analysis.

Person re-identification with block sparse recovery

We consider the problem of automatically re-identifying a person of interest seen in a probe camera view among several candidate people in a gallery camera view. This problem, called person re-identification, is of fundamental importance in several video analytics applications. While extracting knowledge from high-dimensional visual representations based on the notions of sparsity and regularization has been successful for several computer vision problems, such techniques have not been fully exploited in the context of the re-identification problem. Here, we develop a principled algorithm for the re-identification problem in the general framework of learning sparse visual representations. Given a set of feature vectors for a person in one camera view (corresponding to multiple images as they are tracked), we show that a feature vector representing the same person in another view approximately lies in the linear span of this feature set. Furthermore, under certain conditions, the associated coefficient vector can be characterized as being block sparse. This key insight allows us to design an algorithm based on block sparse recovery that achieves state-of-the-art results in multi-shot person re-identification. We also revisit an older feature transformation technique, Fisher discriminant analysis, and show that, when combined with our proposed formulation, it outperforms many sophisticated methods. Additionally, we show that the proposed algorithm is flexible and can be used in conjunction with existing metric learning algorithms, resulting in improved ranking performance. We perform extensive experiments on several publicly available datasets to evaluate the proposed algorithm. A block sparse recovery algorithm for person re-identification is proposed.The algorithm improves the performance of existing metric learning techniques.The algorithm can be used to address generalized ranking problems.

Particle dynamics and multi-channel feature dictionaries for robust visual tracking.

We present a novel approach to solve the visual tracking problem in a particle filter framework based on sparse visual representations. Current state-of-the-art trackers use low-resolution image intensity features in target appearance modeling. Such features often fail to capture sufficient visual information about the target. Here, we demonstrate the efficacy of visually richer representation schemes by employing multi-channel feature dictionaries as part of the appearance model. To further mitigate the tracking drift problem, we propose a novel dynamic adaptive state transition model, taking into account the dynamics of the past states. Finally, we demonstrate the computational tractability of using richer appearance modeling schemes by adaptively pruning candidate particles during each sampling step, and using a fast augmented Lagrangian technique to solve the associated optimization problem. Extensive quantitative evaluations and robustness tests on several challenging video sequences demonstrate that our approach substantially outperforms the state of the art, and achieves stable results.

Learning Affine Hull Representations for Multi-Shot Person Re-Identification

We consider the person re-identification problem, assuming the availability of a sequence of images for each person, commonly referred to as video-based or multi-shot re-identification. We approach this problem from the perspective of learning discriminative distance metric functions. While existing distance metric learning methods typically employ the average feature vector as the data exemplar, this discards the inherent structure of the data. To overcome this issue, we describe the image sequence data using affine hulls. We show that directly computing the distance between the closest points on these affine hulls as in existing recognition algorithms is not sufficiently discriminative in the context of person re-identification. To this end, we incorporate affine hull data modeling into the traditional distance metric learning framework, learning discriminative feature representations directly using affine hulls. We perform extensive experiments on several publicly available data sets to show that the proposed approach improves the performance of existing metric learning algorithms irrespective of the feature space employed to perform metric learning. Furthermore, we advance the state of the art on iLIDS-VID, PRID, and SAIVT, with absolute rank-1 performance improvements of 6.0%, 11.4%, and 6.0% respectively.