Vision–Language Models Empowered Nighttime Object Detection With Consistency Sampler and Hallucination Feature Generator
IEEE Transactions on Image Processing · 2025 · 34 · 8345-8360
DOI: 10.1109/tip.2025.3641316 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Image Processing, 2025. Verfasst von Lihuo He und 6 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 77 Literaturangaben. Gefördert von: National Natural Science Foundation of China.
Abstract (Originalsprache)
Current object detectors often suffer performance degradation when applied to cross-domain scenarios, particularly under challenging visual conditions such as nighttime scenes. This is primarily due to the I3 problems: Inadequate sampling of instance-level features, Indistinguishable feature representation across domains and Inaccurate generation for identical category participation. To address these challenges, we propose a domain-adaptive detection framework that enables robust generalization across different visual domains without introducing any additional inference overhead. The framework comprises three key components. Specifically, the centerness-category consistency sampler alleviates inadequate sampling by selecting representative instance-level features, while the paired centerness consistency loss enforces alignment between classification and localization. Second, VLM-based orthogonality enhancement leverages frozen vision-language encoders with an orthogonal projection loss to improve cross-domain feature distinguishability. Third, hallucination feature generator synthesizes robust instance-level features for missing categories, ensuring balanced category participation across domains. Extensive experiments on multiple datasets covering various domain adaptation and generalization settings demonstrate that our method consistently outperforms state-of-the-art detectors, achieving up to 5.5 mAP improvement, with particularly strong gains in nighttime adaptation.
Förderung
- National Natural Science Foundation of China
Zitieren
APA
He, L., Ke, J., Wang, Z., Li, J., Zhou, K., Wang, Q., & Gao, X. (2025). Vision–Language Models Empowered Nighttime Object Detection With Consistency Sampler and Hallucination Feature Generator. IEEE Transactions on Image Processing, 34, 8345-8360. https://doi.org/10.1109/tip.2025.3641316
Vancouver
He L, Ke J, Wang Z, Li J, Zhou K, Wang Q, et al. Vision–Language Models Empowered Nighttime Object Detection With Consistency Sampler and Hallucination Feature Generator. IEEE Transactions on Image Processing. 2025;34:8345-8360. doi:10.1109/tip.2025.3641316
BibTeX
@article{He2025,
author = {He, Lihuo and Ke, Junjie and Wang, Zhenghao and Li, Jie and Zhou, Kai and Wang, Qi and Gao, Xinbo},
title = {Vision–Language Models Empowered Nighttime Object Detection With Consistency Sampler and Hallucination Feature Generator},
journal = {IEEE Transactions on Image Processing},
year = {2025},
volume = {34},
pages = {8345--8360},
doi = {10.1109/tip.2025.3641316}
}
EUVIMED