Enhancing Person Re-Identification through Tensor Feature Fusion
This addresses the problem of identifying individuals across different camera views for security and surveillance applications, but it appears incremental as it combines existing methods like CNNs, LOMO, GOG, and TXQDA.
The paper tackled person re-identification by developing a system using tensor feature fusion and multilinear subspace learning, achieving effectiveness as demonstrated on VIPeR and PRID450s datasets.
In this paper, we present a novel person reidentification (PRe-ID) system that based on tensor feature representation and multilinear subspace learning. Our approach utilizes pretrained CNNs for high-level feature extraction, along with Local Maximal Occurrence (LOMO) and Gaussian Of Gaussian (GOG ) descriptors. Additionally, Cross-View Quadratic Discriminant Analysis (TXQDA) algorithm is used for multilinear subspace learning, which models the data in a tensor framework to enhance discriminative capabilities. Similarity measure based on Mahalanobis distance is used for matching between training and test pedestrian images. Experimental evaluations on VIPeR and PRID450s datasets demonstrate the effectiveness of our method.