Bi-C2R: Re-index Free Lifelong Person Re-identification
Research Paper#Computer Vision, Person Re-identification, Lifelong Learning🔬 Research|Analyzed: Jan 3, 2026 06:15•
Published: Dec 31, 2025 17:50
•1 min read
•ArXivAnalysis
This paper addresses the challenge of Lifelong Person Re-identification (L-ReID) by introducing a novel task called Re-index Free Lifelong person Re-IDentification (RFL-ReID). The core problem is the incompatibility between query features from updated models and gallery features from older models, especially when re-indexing is not feasible due to privacy or computational constraints. The proposed Bi-C2R framework aims to maintain compatibility between old and new models without re-indexing, making it a significant contribution to the field.
Key Takeaways
- •Addresses the problem of catastrophic forgetting in Lifelong Person Re-identification.
- •Introduces a new task: Re-index Free Lifelong Person Re-identification (RFL-ReID).
- •Proposes the Bi-C2R framework to maintain compatibility between old and new models without re-indexing.
- •Demonstrates leading performance on both RFL-ReID and traditional L-ReID tasks.
Reference / Citation
View Original"The paper proposes a Bidirectional Continuous Compatible Representation (Bi-C2R) framework to continuously update the gallery features extracted by the old model to perform efficient L-ReID in a compatible manner."