Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Videos

DNA: Improving text-based person search through distillation learning, negated relation-aware learning, and augmented

Anh D Nguyen¹, Tam T Ngo², Hoa N Nguyen²

¹Department of Information Systems, VNU University of Engineering and Technology, Vietnam National University at Hanoi, Hanoi, 11314, Vietnam; Graduate School of Software and Information Science, Iwate Prefectural University, Iwate, 0208570, Japan.

Neural Networks : the Official Journal of the International Neural Network Society

|June 26, 2026

Summary

Related Concept Videos

Associative Learning

Associative Learning

Associative learning is a fundamental concept in behavioral psychology, wherein a connection is established between two stimuli or events, leading to a learned response. This process is critical in understanding how behaviors are acquired and modified. Conditioning, the mechanism through which associations are formed, can be divided into two main types: classical conditioning and operant conditioning, each elucidating different aspects of associative learning.
Classical conditioning, also known...

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

<b>A new conspicuously colored species of the genus <i>Hyleoglomeris</i> Verhoeff 1910 (Diplopoda: Glomerida: Glomeridae), with research perspectives on the genus in Japan</b>.

Zootaxa·2026

Same author

<b>Two new earthworm species of the genus <i>Metaphire</i> Sims & Easton, 1972 (Annelida: Oligochaeta: Megascolecidae) from Highlands of Vietnam</b>.

Zootaxa·2026

Same author

The Dallas Consensus for performing and interpreting functional lumen imaging probe panometry: what it means for the clinician.

Current opinion in gastroenterology·2026

Same author

A Pilot Study of Corticosteroid Treatment for Achalasia-Variant Esophagogastric Junction Outflow Obstruction.

Gastro hep advances·2026

Same author

The Combined Utility of the CARS Endoscopic Score and Functional Lumen Imaging Probe (FLIP) Based on Real-Time Assessments.

Neurogastroenterology and motility·2026

Same author

Management of Recurrent Dysphagia after POEM and Heller Myotomy: Current Strategies and Future Directions.

Current gastroenterology reports·2026

This summary is machine-generated.

DNA, a novel framework for text-based person search (TPS), enhances image retrieval by aligning visual and textual data. It effectively models negative expressions and integrates fine-grained details, achieving state-of-the-art results on benchmark datasets.

Area of Science:

Computer Vision
Artificial Intelligence
Natural Language Processing

Background:

Text-based person search (TPS) aims to retrieve individuals from images using text queries.
A key challenge in TPS is effectively aligning visual and textual information in a shared feature space.
Existing methods often struggle with subtle semantic nuances and the effective utilization of negative expressions in textual descriptions.

Purpose of the Study:

To introduce DNA, a novel Text-based Person Search (TPS) framework.
To improve the alignment between visual and textual modalities for more accurate person retrieval.
To address the limitations of current TPS methods by incorporating negative expression modeling and fine-grained feature learning.

Main Methods:

Leveraging distillation-based dual-model learning with a momentum teacher model supervising a CLIP-based student model.

Keywords:

Augmented representation learning Distillation learning Negated relation-aware learning Text-based person search

Related Experiment Videos

Explicitly modeling negative expressions within image captions to capture relational information.

Incorporating negated relation-aware learning and augmented representation learning for discriminative feature alignment.

Main Results:

Achieved state-of-the-art performance on three benchmark datasets: CUHK-PEDES, ICFG-PEDES, and RSTPReID.
Demonstrated superior results compared to recent CLIP-based TPS methods.
Achieved 76.0% Rank-1 and 67.6% mAP on CUHK-PEDES, 67.7% Rank-1 and 40.3% mAP on ICFG-PEDES, and 65.3% Rank-1 and 51.5% mAP on RSTPReID.

Conclusions:

The proposed DNA framework significantly enhances Text-based Person Search performance.
Effective modeling of negative expressions and fine-grained semantics is crucial for improving visual-textual alignment.
DNA offers a robust approach to reducing the visual-textual feature gap and improving inference reliability in person retrieval.