分析图像与文本行人重识别之间的优化冲突
Analyzing the Optimization Conflict Between Image- and Text-Based Person Re-Identification
AI 导读
一篇 arXiv 论文分析图像（I2I）与文本（T2I）行人重识别统一到单一模型时的优化冲突：I2I 追求身份级不变性、要求收缩同身份距离，T2I 依赖实例级文本描述、要求保留该距离，两者对共享视觉编码器提出相反要求，实证表明单个无约束编码器无法同时满足两个目标。
正文 · 原文
Abstract:Unifying image-based (I2I) and text-based (T2I) person re-identification (ReID) within a single model is complicated by a fundamental conflict between their training objectives. While I2I ReID focuses on identity-level invariance across images of the same person captured by multiple cameras, T2I ReID is driven by instance-specific textual descriptions tied to unique visual traits. When both objectives are optimized on a shared vision encoder, they impose opposing requirements on the same embeddings - I2I contracts intra-identity distances while T2I requires them to persist - so improving one task's representation actively degrades the other's. This paper analyzes this optimization conflict and shows empirically that the two objectives are not jointly satisfiable by a single unconstrained encoder. Motivated by this finding, we investigate the conflict through a controlled two-stage pipeline: a shared vision encoder is first trained for I2I invariance, then frozen while only the text encoder is aligned to the fixed visual space, preventing the T2I objective from unwinding the identity-level structure already learned. We conduct extensive experiments across multiple configurations, varying domain-mixing procedures, learning strategies, and task objectives. We observe that I2I pre-training improves subsequent T2I performance, while textual supervision during this pre-training further improves T2I performance, although the vision encoder is frozen during subsequent cross-modal training. We believe these insights provide a meaningful step toward unified ReID systems that reconcile identity-level and instance-level supervision within a single representation.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2606.02242 [cs.CV] |
| (or arXiv:2606.02242v2 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2606.02242 arXiv-issued DOI via DataCite |
Submission history
From: Timur Mamedov [view email]
[v1]
Mon, 1 Jun 2026 13:37:55 UTC (1,355 KB)
[v2]
Fri, 30 Oct 2026 15:42:19 UTC (9,281 KB)
来源：arXiv cs.AI · arxiv.org