Exploiting Feature Correlations by Brownian Statistics for People Detection and Recognition

Bąk, Sławomir; Biagio, Marco San; Kumar, Ratnesh; Murino, Vittorio; Brémond, François

doi:10.1109/tsmc.2016.2531658

Cited by 16 publications

(8 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…By embedding the probability distribution into a matrix, a meta-descriptor such as the covariance descriptor summarizes the local features inside a region. Several methods, such as Entropy and Mutual Information matrix [22], Brownian covariance matrix [23], and covariance matrix on Reproducing Kernel Hilbert Space (RKHS) [24], [25] were proposed to improve the description ability of the covariance descriptor. Gaussian descriptors, such as Shape of Gaussians [14], Global Gaussian [16] and Gaussians Of Local Descriptors (GOLD) [17], embed both the mean vector and covariance matrix of local features.…”

Section: Related Workmentioning

confidence: 99%

Hierarchical Gaussian Descriptors with Application to Person Re-Identification

Matsukawa

Okabe

Suzuki

et al. 2020

IEEE Trans. Pattern Anal. Mach. Intell.

View full text Add to dashboard Cite

Describing the color and textural information of a person image is one of the most crucial aspects of person re-identification (re-id). In this paper, we present novel meta-descriptors based on a hierarchical distribution of pixel features. Although hierarchical covariance descriptors have been successfully applied to image classification, the mean information of pixel features, which is absent from the covariance, tends to be the major discriminative information for person re-id. To solve this problem, we describe a local region in an image via hierarchical Gaussian distribution in which both means and covariances are included in their parameters. More specifically, the region is modeled as a set of multiple Gaussian distributions in which each Gaussian represents the appearance of a local patch. The characteristics of the set of Gaussians are again described by another Gaussian distribution. In both steps, we embed the parameters of the Gaussian into a point of Symmetric Positive Definite (SPD) matrix manifold. By changing the way to handle mean information in this embedding, we develop two hierarchical Gaussian descriptors. Additionally, we develop feature norm normalization methods with the ability to alleviate the biased trends that exist on the descriptors. The experimental results conducted on five public datasets indicate that the proposed descriptors achieve remarkably high performance on person re-id.

show abstract

Section: Related Workmentioning

confidence: 99%

Hierarchical Gaussian Descriptors with Application to Person Re-Identification

Matsukawa

Okabe

Suzuki

et al. 2020

IEEE Trans. Pattern Anal. Mach. Intell.

View full text Add to dashboard Cite

show abstract

“…VRIC 3 [19] proposed this dataset comprising large variations in scale, motion, illumination, occlusion and viewpoint. This 3 Experiment on this dataset is performed post IJCNN conference.…”

Section: Cityflow Datasetmentioning

confidence: 99%

“…Prior to the advancements in deep learning, most embedding learning approaches focus on handcrafting using mixture of multiple feature extractors and/or learning suitable ranking functions to minimize distance across objects of similar identities. Some of the notable approaches are [46,3,31,30,6,23,53].…”

Section: Introductionmentioning

confidence: 99%

Vehicle Re-identification: an Efficient Baseline Using Triplet Embedding

Kuma

Weill

Aghdasi

et al. 2019

2019 International Joint Conference on Neural Networks (IJCNN)

View full text Add to dashboard Cite

1 In this paper we tackle the problem of vehicle reidentification in a camera network utilizing triplet embeddings. Re-identification is the problem of matching appearances of objects across different cameras. With the proliferation of surveillance cameras enabling smart and safer cities, there is an ever-increasing need to re-identify vehicles across cameras. Typical challenges arising in smart city scenarios include variations of viewpoints, illumination and self occlusions. Most successful approaches for reidentification involve (deep) learning an embedding space such that the vehicles of same identities are projected closer to one another, compared to the vehicles representing different identities. Popular loss functions for learning an embedding (space) include contrastive or triplet loss. In this paper we provide an extensive evaluation of these losses applied to vehicle re-identification and demonstrate that using the best practices for learning embeddings outperform most of the previous approaches proposed in the vehicle reidentification literature. Compared to most existing stateof-the-art approaches, our approach is simpler and more straightforward for training utilizing only identity-level annotations, along with one of the smallest published embedding dimensions for efficient inference. Furthermore in this work we introduce a formal evaluation of a triplet sampling variant (batch sample) into the re-identification literature.

show abstract

“…The proposed algorithm is doing well in detecting unusual events in the video surveillance with maintaining a low false alarm rate. However, the last years show introducing some novel algorithms for video object detecting based on statistics [16,17].…”

Section: Literature Reviewmentioning

confidence: 99%

Statistical accuracy analysis of different detecting algorithms for surveillance system in smart city

Al-Yassin

Mousa

Fadhel

et al. 2020

IJEECS

View full text Add to dashboard Cite

<span>Several detecting algorithms are developed for real-time surveillance systems in the smart cities. The most popular algorithms due to its accuracy are: Temporal Differencing, Background Subtraction, and Gaussian Mixture Models. Selecting of which algorithm is the best to be used, based on accuracy, is a good choise, but is not the best. Statistical accuracy anlysis tests are required for achieving a confident decision. This paper presents further analysis of the accuracy by employing four parameters: false recognition, unrecognized, true recognition, and total fragmentation ratios. The results proof that no algorithm is selected as the perfect or suitable for all applications based on the total fragmentation ratio, whereas both false recognition ratio and unrecognized ratio parameters have a significant impact. The mlti-way Analysis of Variate (so-called K-way ANONVA) is used for proofing the results based on SPSS statistics.</span>

show abstract

Exploiting Feature Correlations by Brownian Statistics for People Detection and Recognition

Cited by 16 publications

References 41 publications

Hierarchical Gaussian Descriptors with Application to Person Re-Identification

Hierarchical Gaussian Descriptors with Application to Person Re-Identification

Vehicle Re-identification: an Efficient Baseline Using Triplet Embedding

Statistical accuracy analysis of different detecting algorithms for surveillance system in smart city

Contact Info

Product

Resources

About