Coarse-to-Fine Semantic Alignment: A Hybrid Framework for Forensic Shoeprint Retrieval
IEEE Access · 2026 · 14 · 83389-83405
DOI: 10.1109/access.2026.3695998 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Hyeong Seop Song und 5 weiteren Autor:innen. Die Arbeit nennt 28 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Regional Innovation System and Education (RISE) Program through Gyeongbuk RISE Center, Ministry of Education (MOE) and Gyeongsangbuk-do, Republic of Korea, Basic Science Research Program through the National Research Foundation of Korea.
Abstract (Originalsprache)
While deep learning has advanced image retrieval, conventional unimodal models lack explicit semantic guidance, and existing multimodal paradigms often suffer from a granularity mismatch due to their assumption of global image-caption alignment. This limitation is particularly critical in forensic shoeprint analysis, where distinguishing instances with high global similarity but subtle sub-pattern variations is essential. To address these challenges, we propose a novel hybrid retrieval framework that explicitly bridges the gap between global visual appearances and fine-grained semantic patterns. Unlike a standard architecture, our method integrates visual representations with localized semantic priors by employing a Visual-Centric Query (VCQ) mechanism. This architecture establishes a structured cross-modal alignment that reorients the attention flow, empowering semantically condensed visual tokens to actively probe domain-guided sub-pattern descriptions. This enables the precise token-to-pattern anchoring of visual evidence. By establishing this explicit interaction, the proposed framework not only resolves the granularity mismatch but also ensures interpretability through traceable semantic grounding. Extensive experiments demonstrate significant performance improvements compared to conventional approaches, proving its efficacy as a robust solution for automated crime-scene investigation.
Förderung
- Regional Innovation System and Education (RISE) Program through Gyeongbuk RISE Center
- Ministry of Education (MOE) and Gyeongsangbuk-do, Republic of Korea
- Basic Science Research Program through the National Research Foundation of Korea
- MOE
- Policelab 2.0 (customized research and development for public security sites) Program through Korea Institute of Police Technology
- Korean National Police Agency
Zitieren
APA
Song, H. S., Kim, J. K., Lee, C. Y., Kwon, H. S., Kim, D. J., & Suh, Y. J. (2026). Coarse-to-Fine Semantic Alignment: A Hybrid Framework for Forensic Shoeprint Retrieval. IEEE Access, 14, 83389-83405. https://doi.org/10.1109/access.2026.3695998
Vancouver
Song HS, Kim JK, Lee CY, Kwon HS, Kim DJ, Suh YJ. Coarse-to-Fine Semantic Alignment: A Hybrid Framework for Forensic Shoeprint Retrieval. IEEE Access. 2026;14:83389-83405. doi:10.1109/access.2026.3695998
BibTeX
@article{Song2026,
author = {Song, Hyeong Seop and Kim, Jun Kyu and Lee, Chang Yeop and Kwon, Hyun Seob and Kim, Dong Ju and Suh, Young Joo},
title = {Coarse-to-Fine Semantic Alignment: A Hybrid Framework for Forensic Shoeprint Retrieval},
journal = {IEEE Access},
year = {2026},
volume = {14},
pages = {83389--83405},
doi = {10.1109/access.2026.3695998}
}
EUVIMED