# Continual Representation Learning for Biometric Identification Source: [Continual Representation Learning for Biometric Identification]() Collection index: [llms.txt]() Data updated: 2025-10-27T06:06:39.448Z Authors: Bo Zhao, Shixiang Tang, Dapeng Chen, Hakan Bilen, Rui Zhao Published: 2020-06-08T10:18:06.000Z ## Abstract With the explosion of digital data in recent years, continuously learning new tasks from a stream of data without forgetting previously acquired knowledge has become increasingly important. In this paper, we propose a new continual learning (CL) setting, namely \`\`continual representation learning'', which focuses on learning better representation in a continuous way. We also provide two large-scale multi-step benchmarks for biometric identification, where the visual appearance of different classes are highly relevant. In contrast to requiring the model to recognize more learned classes, we aim to learn feature representation that can be better generalized to not only previously unseen images but also unseen classes/identities. For the new setting, we propose a novel approach that performs the knowledge distillation over a large number of identities by applying the neighbourhood selection and consistency relaxation strategies to improve scalability and flexibility of the continual learning model. We demonstrate that existing CL methods can improve the representation in the new setting, and our method achieves better results than the competitors. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Person Re-identification on CRL-ReID 5-step](): Person Re-Identification; CRL-ReID 5-step; mAP: 51.5. - [Person Re-identification on CRL-ReID (10-step)](): Person Re-Identification; CRL-ReID (10-step); mAP: 44.1. - [Person Re-identification on LReID-Unseen MA-\>SY-\>DU-\>MS-\>CU (generalisation)](): Person Re-Identification; LReID-Unseen MA-\>SY-\>DU-\>MS-\>CU (generalisation); mAP: 39.6. - [Lifelong Person Re-identification on Market1501](): Lifelong Person Re-identification; Market1501; mAP: 58. - [Person Re-identification on DukeMTMC Order-2 (train)](): Person Re-Identification; DukeMTMC Order-2 (train); mAP: 43.5. - [Lifelong Person Re-identification on CUHK-SYSU](): Lifelong Person Re-identification; CUHK-SYSU; mAP: 72.5. - [Lifelong Person Re-identification on CUHK03](): Lifelong Person Re-identification; CUHK03; mAP: 37.4. - [Person Re-identification on Market1501 (train Order-2)](): Person Re-Identification; Market1501 (train Order-2); mAP: 35. - [Person Re-identification on CUHK-SYSU (train Order-2)](): Person Re-Identification; CUHK-SYSU (train Order-2); mAP: 70. - [Lifelong Person Re-identification on DukeMTMC](): Lifelong Person Re-identification; DukeMTMC; mAP: 28.3. - [Lifelong Person Re-identification on MSMT17](): Lifelong Person Re-identification; MSMT17; mAP: 6. - [Person Re-identification on MSMT17 Order-2 (train)](): Person Re-Identification; MSMT17 Order-2 (train); mAP: 4.8. - [Person Re-identification on CUHK03 Order-2 (train)](): Person Re-Identification; CUHK03 Order-2 (train); mAP: 34.5. - [Person Re-identification on Market 1501](): Person Re-Identification; Market 1501; mAP: 67. ## Original sources and implementation [Original paper]() [PDF]()