Multi-shot person re-identification via relational Stein divergence

  title={Multi-shot person re-identification via relational Stein divergence},
  author={Azadeh Alavi and Yan Yang and Mehrtash Tafazzoli Harandi and Conrad Sanderson},
  journal={2013 IEEE International Conference on Image Processing},
  • A. Alavi, Yan Yang, +1 author C. Sanderson
  • Published 15 September 2013
  • Computer Science, Mathematics
  • 2013 IEEE International Conference on Image Processing
Person re-identification is particularly challenging due to significant appearance changes across separate camera views. In order to re-identify people, a representative human signature should effectively handle differences in illumination, pose and camera parameters. While general appearance-based methods are modelled in Euclidean spaces, it has been argued that some applications in image and video analysis are better modelled via non-Euclidean manifold geometry. To this end, recent approaches… 
Image Analysis on Symmetric Positive Definite Manifolds
Over the last two decades, the research community has witnessed extensive research growth in the field of analysing and understanding scenes. Automatic scene analysis can support many critical
Adaptive Metric Learning for People Re-Identification
An adaptive learning method to learn an adaptive distance metric is proposed, which integrates prior knowledge learned from a large existing auxiliary dataset and task-specific information extracted from a much smaller training dataset.
Person Reidentification Using Multiple Egocentric Views
A person re-id framework designed for a network of multiple wearable devices is presented and an online scheme is proposed as a direct extension of the batch method to ensure its utility in practical applications where a large amount of observations are available every instant.
Consistent Re-identification in a Camera Network
A network consistent re-identification (NCR) framework is proposed, which is formulated as an optimization problem that not only maintains consistency in re- identification results across the network, but also improves the camera pairwise re-Identification performance between all the individual camera pairs.
Multiple-Shot People Re-Identification by Patch-Wise Learning
The proposedpatch-wise learning based approach to deal with the multiple-shot people re-identification task is formulated as a patch-wise set-toset matching problem, with each patch set being matched using a specifically learned Mahalanobis distance metric.
Person re-identification using multiple first-person-views on wearable devices
This framework utilizes a global data association method termed as Network Consistent Reidentification (NCR) that not only helps in maintaining consistency in association results across the network, but also improves the pair-wise face re-identification accuracy.
Reranking optimization for person re-identification under temporal-spatial information and common network consistency constraints
A re-ranking optimization framework under temporal-spatial information and common network consistency constraints is proposed to compensate the accuracy deterioration caused by appearance only based ReID methods in all pairwise cameras.
Analysis of Skeletal Shape Trajectories for Person Re-Identification
The main contribution of this work resides in the introduced distance that combines temporal information as well as global and local geometrical ones in the motion dynamic of human motion.
Efficient clustering on Riemannian manifolds: A kernelised random projection approach
This work proposes a kernelised random projection framework for clustering manifold points via kernel space, which can preserve the geometric structure of the original space, but is computationally efficient.
Joint Temporal-Spatial Information and Common Network Consistency Constraint for Person Re-identification
A joint temporal-spatial information and common network consistency constraint framework is proposed to improve the re-identification performance in all pairwise cameras and results validate that the proposed framework obtains significantly better performance compared to the state of the art camera network person re-Identification methods.


Relational divergence based classification on Riemannian manifolds
Experiments on face recognition, person re-identification and texture classification show that the proposed method outperforms state-of-the-art approaches, such as Tensor Sparse Coding, Histogram Plus Epitome and the recent Riemannian Locality Preserving Projection.
Dirichlet process mixture models on symmetric positive definite matrices for appearance clustering in video surveillance applications
A novel application of the Dirich-let Process Mixture Model framework is proposed towards unsupervised clustering of symmetric positive definite matrices, extending the existing K-means type clustering algorithms based on the logdet divergence measure and derive the counterpart of it in a Bayesian framework, which leads to the Wishart-Inverse Wishart conjugate pair.
Kernel analysis over Riemannian manifolds for visual recognition of actions, pedestrians and textures
Experiments on several visual classification tasks show that the proposed embedding into the Reproducing Kernel Hilbert Space by introducing a Riemannian pseudo kernel obtains considerable improvements in discrimination accuracy.
K-tangent spaces on Riemannian manifolds for improved pedestrian detection
A general discriminative model based on the combination of several tangent spaces is proposed in order to preserve more details of the structure of the Riemannian structure.
Person re-identification by symmetry-driven accumulation of local features
In this paper, we present an appearance-based method for person re-identification. It consists in the extraction of features that model three complementary aspects of the human appearance: the
Tangent Bundles on Special Manifolds for Action Recognition
  • Y. Lui
  • Mathematics, Computer Science
    IEEE Transactions on Circuits and Systems for Video Technology
  • 2012
The empirical results reveal that the proposed paradigm frames the classification scheme as a nearest neighbor based on the intrinsic distance, prior training is unnecessary and the method is highly competitive with the current state-of-the-art methods, robust to small alignment errors, and yet simpler.
Efficient similarity search for covariance matrices via the Jensen-Bregman LogDet Divergence
A novel dissimilarity measure for covariance matrices is advocated: the Jensen-Bregman LogDet Divergence, which enjoys several useful theoretical properties, but its greatest benefits are lower computational costs (compared to standard approaches) and amenability for use in nearest-neighbor retrieval.
Statistical Computations on Grassmann and Stiefel Manifolds for Image and Video-Based Recognition
This paper discusses how commonly used parametric models for videos and image sets can be described using the unified framework of Grassmann and Stiefel manifolds, and derives statistical modeling of inter and intraclass variations that respect the geometry of the space.
Multiple-Shot Person Re-identification by HPE Signature
In this paper, we propose a novel appearance-based method for person re-identification, that condenses a set of frames of the same individual into a highly informative signature, called Histogram
Sparse Coding and Dictionary Learning for Symmetric Positive Definite Matrices: A Kernel Approach
Experiments show that the proposed sparse coding approach achieves notable improvements in discrimination accuracy, in comparison to state-of-the-art methods such as tensor sparse coding, Riemannian locality preserving projection, and symmetry-driven accumulation of local features.