Person Re-identification by Descriptive and Discriminative Classification

  title={Person Re-identification by Descriptive and Discriminative Classification},
  author={Martin Hirzer and Csaba Beleznai and Peter M. Roth and Horst Bischof},
Person re-identification, i.e., recognizing a single person across spatially disjoint cameras, is an important task in visual surveillance. [] Key Method First, given a specific query, we rank all samples according to a feature-based similarity, where appearance is modeled by a set of region covariance descriptors. Next, a discriminative model is learned using boosting for feature selection, which provides a more specific classifier. The proposed approach is demonstrated on two datasets, where we show that…

Combining Descriptive and Discriminative Information for Person Re-Identification

This thesis introduces an application-focused approach of integrating a descriptive and a discriminative person model into a single system and addresses metric learning, a relatively new direction in the field of person re-identification.

Person Re-Identification Using Hybrid Representation Reinforced by Metric Learning

The experimental results show that the hybrid hand-crafted and deep features outperformed the existing state-of-the-art in approaches in the unsupervised paradigm.

Person re-identification by combining features in a learning based framework

This study investigates the discriminative ability of different features extracted from an image in a binary classification framework, and proposes a learning based method to combine HSV histogram, Maximally Stable Colour Regions (MSCR) and Speeded-Up Robust Features (SURF) distances in a single framework.

Weighted hybrid features for person re-identification

This paper proposes an effective new person reidentification model which incorporates several recent state-of-the-art feature extraction methodologies such as GOG, WHOS and LOMO features into a single framework and is tested on multiple benchmark person re-identification datasets where it outperforms many other state of theart methodologies.

Person Re-Identification with Discriminatively Trained Viewpoint Invariant Dictionaries

This paper proposes to learn a dictionary that is capable of discriminatively and sparsely encoding features representing different people, and directly addresses two key challenges in person re-identification: viewpoint variations and discriminability.

Exploiting Multiple Detections for Person Re-Identification

This paper presents a CWBTF framework for the task of transforming appearance from one camera to another, and a re-identification framework where the pedestrian images are segmented into meaningful parts and features are extracted from such parts, as well as from the whole body.

Person Re-identification Based on Adaptive Feature Selection

A person re-identification approach by using adaptive feature selection method that detects human and human body parts, extracts certain features on each part adaptively driven by their unique and inherent appearance attributes, and finds out the similarity between corresponding body parts.

Spatial Pyramid-Based Statistical Features for Person Re-Identification: A Comprehensive Evaluation

This paper proposes a spatial pyramid-based statistical feature extraction framework as a unified pipeline of feature extraction and combination for person Re-Id, and systematically evaluate the configuration details infeature extraction and the fusion strategies in feature combination.

Person Re-Identification by Discriminative Selection in Video Ranking

This work presents a novel model to automatically select the most discriminative video fragments from noisy/incomplete image sequences of people from which reliable space-time and appearance features can be computed, whilst simultaneously learning a video ranking function for person ReID.



Person Re-Identification by Support Vector Ranking

This work converts the person re-identification problem from an absolute scoring p roblem to a relative ranking problem and develops an novel Ensemble RankSVM to overcome the scalability limitation problem suffered by existing SVM-based ranking methods.

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

Person Reidentification Using Spatiotemporal Appearance

A novel spatiotemporal segmentation algorithm is employed to generate salient edgels that are robust to changes in appearance of clothing and invariant signatures are generated by combining normalized color and salient edgel histograms.

Person Re-identification Using Haar-based and DCD-based Signature

Two approaches for person re-identification problem are presented based onhaar-like features and dominant color descriptors and the AdaBoostscheme is applied to both descriptors to achieve invariant and discriminative signature.

Learning Discriminative Appearance-Based Models Using Partial Least Squares

  • W. R. SchwartzL. Davis
  • Computer Science
    2009 XXII Brazilian Symposium on Computer Graphics and Image Processing
  • 2009
The experimental results demonstrate that the use of an enriched feature set analyzed by PLS reduces the ambiguity among different appearances and provides higher recognition rates when compared to other machine learning techniques.

Viewpoint Invariant Pedestrian Recognition with an Ensemble of Localized Features

It is shown how both an object class specific representation and a discriminative recognition model can be learned using the AdaBoost algorithm, which allows many different kinds of simple features to be combined into a single similarity function.

Learning Pairwise Dissimilarity Profiles for Appearance Recognition in Visual Surveillance

A dissimilarity distance measure is introduced and linearly or nonlinearly combine it with direct distances to improve the scalability of classifiers to larger number of categories.

A Boundary-Fragment-Model for Object Detection

The BFM detector is able to represent and detect object classes principally defined by their shape, rather than their appearance, and to achieve this with less supervision (such as the number of training images).

Appearance modeling for tracking in multiple non-overlapping cameras

  • O. JavedK. ShafiqueM. Shah
  • Computer Science
    2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05)
  • 2005
It is shown that all brightness transfer functions from a given camera to another camera lie in a low dimensional subspace and it is demonstrated that this subspace can be used to compute appearance similarity.

Learning object detection from a small number of examples: the importance of good features

  • K. LeviYair Weiss
  • Computer Science
    Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2004. CVPR 2004.
  • 2004
This work shows that using local edge orientation histograms (EOH) as features can significantly improve performance compared to the standard linear features used in existing systems and enables learning a system that seems to outperform the state of the art in real-time systems even with a small number of training examples.