Person Reidentification Using Facial and Fabric Feature Extraction
Keywords:
Bag of features, Feature matching, Pedestrian detection, Person Re-id, Spatial pyramid matchingAbstract
Purpose: Person identification across surveillance cameras with non-overlapping fields of view has emerged as a highly challenging and intriguing area of study in smart surveillance. Though numerous approaches have been carried out, proposed, and developed, unresolved issues and limitations persist.
Methods: Existing re-identification methodologies generally, involve extracting feature vectors from images or video frames and applying various similarity or dissimilarity measures to compare these vectors. Some approaches rely on models based on fabric colour or facial features information, the ultimate objective of all these methods is to attain higher matching accuracy and at the same time lower computational costs. This work involves the study of person re-identification, using the classification of feature vectors for facial and fabric images, represented as a visual word dictionary, and person matching is carried out using different baseline methods. The primary novelty of this work lies in a selective and context-aware fusion of facial and fabric features within a Bag of Visual Words (BoVW) framework enhanced by Spatial Pyramid Matching (SPM) for person re-identification in non-overlapping camera environments. Unlike existing approaches that rely solely on either global appearance features or deep learning–based representations, the author's method explicitly separates and independently models facial and fabric regions, enabling robust re-identification under clothing variations and pose changes.
Results: Experiments are done with various benchmark datasets, and comparisons are done with various baseline methods, which show that the method gives 99% accuracy for the ETHZ dataset.
Conclusion: Person ReID using facial and fabric feature extraction has been carried out using Bag of Visual Words (BoVW) features. Also, to include spatial information of features, BoVW features are grouped using spatial pyramid matching, and the performance of the system has been improved.
References
O. Hamdoun, F. Moutarde, B. Stanciulescu, and B. Steux, “Person re-identification in multi-camera system by signature based on interest point descriptors collected on short video sequences,” 2008 Second ACM/IEEE International Conference on Distributed Smart Cameras, Sep. 2008.
N. Gheissari, T. B. Sebastian, and R. Hartley, “Person Reidentification Using Spatiotemporal Appearance,” Computer Vision and Pattern Recognition, Jun. 2006
F. Porikli, O. Tuzel, and P. Meer, “Covariance Tracking using Model Update Based on Lie Algebra,” IEEE Conf. Computer Vision and Pattern Recognition, Jul. 2006.
X. Wang, G. Doretto, T. Sebastian, J. Rittscher, and P.-F. Tu, “Shape and Appearance Context Modelling,” International Conference on Computer Vision, Dec. 2007.
D. Gray, S. Brennan, and H. Tao, “Evaluating appearance models for recognition, reacquisition, and tracking,” Jan. 01, 2007
Z. Lin and L. S. Davis, “Learning Pairwise Dissimilarity Profiles for Appearance Recognition in Visual Surveillance,” Lecture Notes in Computer Science, pp. 23–34, Jan. 2008.
Q. Li, X. Peng, Y. Qiao, and Q. Hao, “Unsupervised person re-identification with multi-label learning guided self-paced clustering,” Pattern Recognition, vol. 125, p. 108521, May 2022.
D. Cheng, Y. Gong, S. Zhou, J.-J. Wang, and N. Zheng, “Person Re-identification by Multi-Channel Parts-Based CNN with Improved Triplet Loss Function,” Computer Vision and Pattern Recognition, Jun. 2016.
W.-S. Zheng, S. Gong, and T. Xiang, “Reidentification by Relative Distance Comparison,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 35, no. 3, pp. 653–668, Mar. 2013.
S.-Z. Chen, C.-C. Guo, and J.-H. Lai, “Deep Ranking for Person Re-Identification via Joint Representation Learning,” IEEE Transactions on Image Processing, vol. 25, no. 5, pp. 2353–2367, May 2016.
L. Bazzani, M. Cristani, A. Perina, and V. Murino, “Multiple-shot person re-identification by chromatic and epitomic analyses,” Pattern Recognition Letters, vol. 33, no. 7, pp. 898–903, May 2012.
J. Zhang, L. Niu, and L. Zhang, “Person Re-Identification with Reinforced Attribute Attention Selection,” IEEE Transactions on Image Processing, vol. 30, pp. 603–616, 2021
D. Chen, S. Zhang, W. Ouyang, J. Yang, and Y. Tai, “Person Search by Separated Modelling and A Mask-Guided Two-Stream CNN Model,” IEEE Transactions on Image Processing, vol. 29, pp. 4669–4682, Jan. 2020.
M. Ye and P. C. Yuen, “PurifyNet: A Robust Person Re-Identification Model with Noisy Labels,” IEEE Transactions on Information Forensics and Security, vol. 15, pp. 2655–2666, Jan. 2020.
L. Zhang, G. Du, F. Liu, H. Tu, and X. Shu, “Global-Local Multiple Granularity Learning for Cross-Modality Visible-Infrared Person Reidentification,” IEEE Transactions on Neural Networks and Learning Systems, pp. 1–11, Jan. 2021.
H. Wang, L. Jiao, S. Yang, L. Li, and Z. Wang, “Simple and Effective: Spatial Rescaling for Person Reidentification,” IEEE Transactions on Neural Networks and Learning Systems, vol. 33, no. 1, pp. 145–156, Jan. 2022.
M. Fu et al., “Improving Person Re-identification using A Self-focusing Network in Internet of Things,” IEEE Internet of Things Journal, vol. 9, no. 12, pp. 1–1, 2021.
S. Zhong, Z. Bao, S. Gong, and K. Xia, “Person Reidentification Based on Pose-Invariant Feature and B-KNN Reranking,” IEEE Transactions on Computational Social Systems, vol. 8, no. 5, pp. 1272–1281, Mar. 2021.
