ReText: Text Boosts Generalization in Image-Based Person Re-identification

Timur Mamedov; Karina Kvanchiani; Anton Konushin; Vadim Konushin

cs.CVFeb 5, 2026

ReText: Text Boosts Generalization in Image-Based Person Re-identification

Authors: Timur Mamedov, Karina Kvanchiani, Anton Konushin, Vadim Konushin

Organizations: Tevian, Moscow, Russia · Lomonosov Moscow State University

Abstract

Generalizable image-based person re-identification (Re-ID) aims to recognize individuals across cameras in unseen domains without retraining. While visual features are sensitive to domain shifts, natural language descriptions remain consistent across domains, regardless of lighting changes and other biases. Since well-known multi-camera datasets lack textual descriptions, we update them with captions through a fine-tuned Vision-Language Model (VLM). We propose ReText, a novel method trained with textual supervision to enrich visual features with semantic cues. Besides, inspired by recent findings showing that incorporating single-camera data into training improves generalization, we also adopt this strategy. However, in contrast to such prior works, we exploit not only visual information but also the textual annotations naturally accompanying these samples. Finally, during training, ReText jointly optimizes three tasks: (1) Re-ID, (2) image-text matching, and (3) image reconstruction guided by text. Experiments demonstrate that ReText achieves strong generalization and significantly outperforms state-of-the-art methods on cross-domain Re-ID benchmarks. The expanded multi-camera datasets and VLM weights are publicly available at https://github.com/timmzak/ReText.

Figures & tables

Figure 1: Visual comparison of attention maps for models trained with and without textual supervision. The text-supervised model tends to focus on discriminative attributes like clothing and accessories instead of spreading attention over less informative areas. Attention maps are obtained from the last-layer [CLS]-to-patch attention, averaged across heads, and min-max normalized.

Figure 2: Overview of ReText. The model receives three types of inputs: (a) images from multi-camera, single-camera, or both data sources, (b) masked versions of images, and (c) textual captions. It jointly optimizes the Re-ID loss, the image-text matching loss, and the text-guided reconstruction loss.

Dataset	Multi	Text	#Images	#IDs	#Cams
CUHK03-NP	✓	✗	$14,097$	$1,467$	$2$
Market-1501	✓	✗	$32,217$	$1,501$	$6$
MSMT17	✓	✗	$126,441$	$4,101$	$15$
CUHK-SYSU	✓	✗	$34,574$	$11,934$	—
RandPerson	✓	✗	$132,145$	$8,000$	$19$
SYNTH-PEDES	✗	✓	$4,791,771$	$312,321$	—

Table 1: Statistics of datasets used in our experiments. The single-camera SYNTH-PEDES is significantly larger than all multi-camera datasets and textually annotated.

Figure 3: The block-structured similarity matrix. The caption embeddings are obtained from the text encoder trained within ReText on MSMT17 with textual supervision.

Panel of Table 2 — Table 2: Effect of single-camera data and textual supervision on cross-domain Re-ID. MSMT17 is used as multi-camera data for training.

Method	Tr. D.	CUHK03-NP		Market-1501		MSMT17		Average
Method	Tr. D.	$Rank_{1}$	$mAP$	$Rank_{1}$	$mAP$	$Rank_{1}$	$mAP$	$Rank_{1}$	$mAP$
TransMatcher [ 21 ]	Market-1501	$22.2$	$21.4$	—	—	$47.3$	$18.4$	$34.8$	$19.9$
QAConv-GS [ 22 ]		$19.1$	$18.1$	$91.6$	$75.5$	$45.9$	$17.2$	$32.5$	$17.6$
PAT [ 27 ]		$25.4$	$26.0$	$92.4$	$81.5$	$42.8$	$18.2$	$34.1$	$22.1$
LDU [ 30 ]		$18.5$	$18.2$	—	—	$35.7$	$13.5$	$27.1$	$15.8$
DCAC [ 14 ]		$33.2$	$32.5$	$94.9$	$86.8$	$52.1$	$23.4$	$42.7$	$27.9$

Table 3: Comparison of ReText with other state-of-the-art methods. Tr. D. denotes the training dataset. Metrics on the Market-1501 and MSMT17 datasets are excluded from average values in the first and second blocks, respectively.

Image-Based Re-ID with Text Tokens
Method	Market-1501		MSMT17		Average
Method	$Rank_{1}$	$mAP$	$Rank_{1}$	$mAP$	$Rank_{1}$	$mAP$
CLIP-ReID [ 18 ]	$95.4$	$90.5$	$89.7$	$75.8$	$92.6$	$83.2$
CLIP-DFGS [ 42 ]	$93.7$	$85.5$	$86.4$	$66.3$	$90.1$	$75.9$
Unified Re-ID Methods
PLIP [ 48 ]	$\underline{97.3}$	$93.2$	—	—	—	—

Table 4: Comparison of ReText with other methods that support language supervision through text tokens or by unifying methods for image and text queries. STL and MTL versions of Instruct-ReID mean single-task and multi-task learning, respectively.

Table 8

Dataset	Texts	CUHK03-NP		Market-1501		Average
Dataset	Texts	$Rank_{1}$	$mAP$	$Rank_{1}$	$mAP$	$Rank_{1}$	$mAP$
MSMT17	32,621	$61.9$	$61.8$	$93.3$	$81.4$	$77.6$	$71.6$
SYNTH-PEDES	32,621	$61.4$	$62.3$	$93.3$	$82.5$	$77.4$	$72.4$
	100,000	$61.5$	$62.6$	$92.7$	$82.1$	$77.1$	$72.4$
	1,000,000	$\underline{62.6}$	$\underline{62.8}$	$93.0$	$\underline{82.7}$	$\underline{77.8}$	$\underline{72.8}$
	2,000,000	$61.9$	$62.7$	$\underline{93.2}$	$82.4$	$77.6$	$72.6$

Table 7: Impact of increasing the quantity of single-camera data. The model is trained on MSMT17 as multi-camera data in both Mt (first row) and MSt (other rows) setups.

Configuration	Text	CUHK03-NP		Market-1501
Configuration	Text	$Rank_{1}$	$mAP$	$Rank_{1}$	$mAP$
in all Re-ID losses	✗	$49.9$	$51.3$	$92.4$	$78.2$
in all Re-ID losses	✓	$63.4$	$62.9$	$93.3$	$82.8$
in no Re-ID losses	✓	$63.4$	$\mathbf{63.1}$	$\mathbf{93.6}$	$\mathbf{83.6}$
only in $\mathcal{L}_{ins}$	✓	$63.1$	$62.9$	$\mathbf{93.6}$	$\mathbf{83.6}$
only in $\mathcal{L}_{aug}$	✓	$62.6$	$62.3$	$93.5$	$83.3$

Table 8: Impact of using single-camera data in different Re-ID loss functions. The model is trained on MSMT17.

Loss	$K_{s}$	$Rank_{1}$	$mAP$
CLIP loss	$1$	$59.8$	$60.3$
	$2$	$59.6$	$60.1$
	$4$	$59.5$	$59.2$
Soft CLIP loss	$1$	$59.8$	$60.3$
	$2$	$60.0$	$60.5$
	$4$	$59.0$	$59.5$

Table 9: Analysis of the number of same-identity instances (

K_{s}

) from single-camera data in a mini-batch for different image-text matching loss functions. The model is trained on MSMT17 and tested on CUHK03-NP.

Table 10: a) Analysis of the number of single-camera images in a mini-batch (

N_{s}

). b) Analysis of the

\alpha

parameter in Eq. 5 for the Identity-aware Matching Loss. c) Analysis of the temperature parameter

\tau

in Eq. 8 for the Structure-preserving Loss. The model is trained on MSMT17 and tested on CUHK03-NP.

Figure 4: Examples of image reconstruction in ReText. Each example shows the original image (left), masked input (middle), and the reconstructed output (right), alongside the corresponding caption. The model successfully recovers key visual details, highlighting the effectiveness of image-text training.

	Method	M+MS+CS $\rightarrow$ C3		M+CS+C3 $\rightarrow$ MS		MS+CS+C3 $\rightarrow$ M		Average
	Method	$Rank_{1}$	$mAP$	$Rank_{1}$	$mAP$	$Rank_{1}$	$mAP$	$Rank_{1}$	$mAP$
Protocol 2	META [ 39 ]	$35.1$	$36.3$	$49.9$	$22.5$	$86.1$	$67.5$	$57.0$	$42.1$
	ACL [ 41 ]	$41.8$	$41.2$	$45.9$	$20.4$	$89.3$	$74.3$	$59.0$	$45.3$
	CLIP-ReID [ 18 ]	$41.9$	$42.1$	$53.1$	$26.6$	$84.4$	$68.8$	$59.8$	$45.8$
	ReFID [ 29 ]	$34.8$	$33.3$	$39.8$	$18.3$	$85.3$	$67.6$	$53.3$	$39.7$
	GMN [ 32 ]	$42.1$	$43.2$	$50.9$	$24.4$	$87.1$	$72.3$	$60.0$	$46.6$

Table 11: Comparison of ReText with other state-of-the-art methods according to Protocols 2 and 3. In this table, C3 is CUHK03-NP, MS is MSMT17, M is Market-1501, and CS is CUHK-SYSU. ReMix and DynaMix are excluded from this comparison because its authors did not evaluate the method using these protocols.

Method	Tr. D.	CUHK03-NP		Market-1501		Average
Method	Tr. D.	$Rank_{1}$	$mAP$	$Rank_{1}$	$mAP$	$Rank_{1}$	$mAP$
QAConv [ 20 ]	MSMT17-merged	$25.3$	$22.6$	$72.6$	$43.1$	$49.0$	$32.9$
TransMatcher [ 21 ]		$31.9$	$30.7$	$82.6$	$58.4$	$57.3$	$44.6$
QAConv-GS [ 22 ]		$27.6$	$28.0$	$82.4$	$56.9$	$55.0$	$42.5$
PAT [ 27 ]		$27.4$	$28.7$	$72.8$	$48.6$	$50.1$	$38.7$
CLIP-DFGS [ 42 ]		$37.1$	$25.7$	$80.9$	$55.2$	$59.0$	$40.5$

Table 12: Comparison of ReText with other state-of-the-art methods according to Protocol 1 using the MSMT17-merged training dataset.

LLaVA	CUHK03-NP		Market-1501		Average
LLaVA	$Rank_{1}$	$mAP$	$Rank_{1}$	$mAP$	$Rank_{1}$	$mAP$
HAM [ 12 ]	$\underline{60.0}$	$\underline{61.1}$	$\mathbf{93.6}$	$\underline{81.4}$	$\underline{76.8}$	$\underline{71.2}$
Ours (CIR)	$59.7$	$60.7$	$\underline{93.5}$	$\underline{81.4}$	$76.6$	$71.1$
Ours (UFine3C)	$\mathbf{61.9}$	$\mathbf{61.8}$	$93.3$	$\underline{81.4}$	$\mathbf{77.6}$	$\mathbf{71.6}$
Ours (UFine3C † )	$59.7$	$60.6$	$\mathbf{93.6}$	$\mathbf{81.5}$	$76.7$	$71.0$

Table 13: Comparison of LLaVA models fine-tuned on different datasets. The CIR is the combination of CUHK-PEDES, ICFG-PEDES and RSTPReid datasets. UFine3C † denotes LLaVA fine-tuned on UFine3C, excluding images overlapping with the multi-camera datasets. The model is trained on MSMT17.

Dataset	#Texts	#Texts w. reps.	#Too Long Texts
CUHK03-NP	$14,097$	$1$	$0$
Market-1501	$32,217$	$15$	$4$
MSMT17	$126,441$	$127$	$16$
CUHK-SYSU	$34,574$	$174$	$5$
RandPerson	$132,145$	$350$	$53$

Table 14: Number of generated captions per dataset, and counts of captions containing repeated phrases (#Texts w. reps.) and exceeding the length threshold (#Too Long Texts).

Figure 5: Examples of defective generated captions. Hallucinated attributes that are not supported by the image are shown in red text, while repeated phrases produced by degenerate looping in the generation process are shown in gray text.

Figure 6: Qualitative examples of the attribute-matching categories from Figure 7 , showing manually annotated (M.A.) text, generated (Gen.) text, and their corresponding extracted attribute sets. The fields that differ between M.A. and Gen. texts are highlighted in red.

Figure 7: Attribute-level agreement between generated and human-annotated captions. Attributes were manually extracted from both captions and matched. Each chart displays the share of matches (correctly reproduced), extra attributes (noted in the generated text but omitted in the original), missing attributes (present in the original but omitted in the generated), and incorrect values (conflicting values).

Figure 8: High-frequency words of MSMT17 generated captions.

Figure 9: Examples of the textual descriptions generated by LLaVA for the images in the MSMT17 dataset.

Figure 10: Distribution of the word count in each generated caption for CUHK03, MSMT17, Market-1501, RandPerson, and CUHK-SYSU datasets. Min., max., avg. corresponds to the minimum, maximum, and average number of words.

Person Re-Identification

Person re-identification matches the same individual across camera views, times, clothing, and sensing modalities despite changes in appearance.

See all 84 papers

Prompt-Anchored Vision-Text Distillation for Lifelong Person Re-identification
May 6, 2026Wen Wen, Hao Chen, Shiliang ZhangCross-Modal Knowledge Distillation Continual Learning
University of Electronic Science and Technology of China, China · Harbin Institute of Technology, Shenzhen, China · Peking University, China
Multi-Grained Vision-Language Alignment for Domain Generalized Person Re-Identification
Mar 14, 2026Jiachen Li, Xiaojin Gong, Dongping ZhangPerson Re-Identification Vision-Language Alignment
College of Information Science and Electronic Engineering, Zhejiang University · College of Information Engineering, China Jiliang University
DiffReID: Discriminative Diffusion Model for Object Re-Identification
Sep 29, 2026Yingquan Wang, Pingping Zhang, Dong Wang +1Representation Learning Query Re-Identification
School of Information and Communication Engineering, Dalian University of Technology · School of Future Technology, Dalian University of Technology

May 6, 2026J/K move · Enter open · S save

Prompt-Anchored Vision-Text Distillation for Lifelong Person Re-identification

Wen Wen, Hao Chen, Shiliang Zhang

University of Electronic Science and Technology of China, China · Harbin Institute of Technology, Shenzhen, China · Peking University, China

Cross-Modal Knowledge Distillation Continual Learning Vision-Language Alignment Visual Prompt Learning Person Re-Identification

Open paper AskPDF arXiv

Lifelong person re-identification (LReID) aims to train a generalizable model with sequentially collected data. However, such models often suffer from semantic drift, limited adaptability, and catastrophic forgetting as new domains emerge. Existing exemplar-free approaches largely rely on visual-only distillation or parameter regularization, while overlooking the potential of auxiliary modalities, such as text, to preserve semantic stability and enable incremental plasticity. We observe that the frozen text encoder in pretrained vision-language models can serve as a stable semantic anchor across domains. To decouple the roles of vision and text, we propose Prompt-Anchored vision-text Distillation (PAD), an asymmetric vision-text framework for semantic alignment and cross-domain generalization. On the textual side, we distill prompts to preserve vision-text alignment under a fixed semantic space, acting as a global semantic reference rather than a dominant learning signal. On the visual side, an EMA-based teacher with an adaptive prompt pool enables domain-wise adaptation by allocating new slots while freezing past ones. Extensive experiments show that PAD substantially outperforms state-of-the-art methods across seen and unseen domains, achieving a strong balance between stability and plasticity. Project page is available at https://github.com/zu-zi/PAD.

Abstract

Person Re-Identification

Explore similar work

Prompt-Anchored Vision-Text Distillation for Lifelong Person Re-identification

Multi-Grained Vision-Language Alignment for Domain Generalized Person Re-Identification

DiffReID: Discriminative Diffusion Model for Object Re-Identification