DR-CLIP: A Deformable Vision–Language Model for Scale-Invariant Object Counting in Remote Sensing Images
Sensors · 16.03.2026 · 26 (6) · 1863
DOI: 10.3390/s26061863 ↗Kurz gefasst
Fachartikel in der Zeitschrift Sensors, online erschienen am 16.03.2026. Verfasst von Jingzhe Nie und 4 weiteren Autor:innen. Beteiligte Einrichtung: College of Information Science and Engineering, Shandong Agricultural University, Tai'an 271018, China. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 59 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: College Student Innovation and Entrepreneurship Training Program, National Natural Science Foundation of China, Natural Science Foundation of Shandong Province.
Abstract (Originalsprache)
Object counting in remote sensing images is valuable for applications such as urban planning and environmental monitoring. However, it remains challenging due to heterogeneous annotations, semantic ambiguity in open-vocabulary queries, and performance degradation of small targets. To address these limitations, we propose DR-CLIP (Deformable Remote CLIP), a vision–language model for remote sensing image counting that incorporates deformable visual feature extraction with text-guided prediction. DR-CLIP includes a (1) Region-to-Instruction (R2I) mechanism to convert points, bounding boxes, and polygons into a unified image–text training representation, a (2) Multi-scale Deformable Attention (MSDA) to enhance discriminative feature extraction across extreme scale variations and cluttered backgrounds, and a (3) Text-Guided Counting Head that establishes robust cross-modal alignment through contrastive learning, achieving open-vocabulary counting capability without category-specific retraining. On DOTA-v2.0, DR-CLIP achieves a Mean Absolute Error (MAE) of 2.34 and a Root Mean Squared Error (RMSE) of 3.89, outperforming baselines by 19.0% in MAE. The MSDA module significantly increases Small-Object Recall (SOR) to 0.824, which is especially effective in situations involving dense and small object counting. In cross-modal retrieval, DR-CLIP attains R@1 scores of 68.3% (image-to-text) and 72.1% (text-to-image) on the Remote Sensing Image Captioning Dataset (RSICD). The framework generalizes robustly, with only 8.7% performance degradation in cross-domain tests, which is significantly lower than the 23.4% drop observed in baseline methods.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- College of Information Science and Engineering, Shandong Agricultural University, Tai'an 271018, China.
ORCID-Profile: Jingzhe Nie, Xu Lu
Förderung
- College Student Innovation and Entrepreneurship Training Program
- National Natural Science Foundation of China
- Natural Science Foundation of Shandong Province
Zitieren
APA
Nie, J., Liu, Q., Li, T., Lu, X., & Zhang, L. (2026). DR-CLIP: A Deformable Vision–Language Model for Scale-Invariant Object Counting in Remote Sensing Images. Sensors, 26(6), 1863. https://doi.org/10.3390/s26061863
Vancouver
Nie J, Liu Q, Li T, Lu X, Zhang L. DR-CLIP: A Deformable Vision–Language Model for Scale-Invariant Object Counting in Remote Sensing Images. Sensors. 2026;26(6):1863. doi:10.3390/s26061863
BibTeX
@article{Nie2026,
author = {Nie, Jingzhe and Liu, Qun and Li, Tianze and Lu, Xu and Zhang, Liang},
title = {DR-CLIP: A Deformable Vision–Language Model for Scale-Invariant Object Counting in Remote Sensing Images},
journal = {Sensors},
year = {2026},
volume = {26},
number = {6},
pages = {1863},
doi = {10.3390/s26061863}
}
EUVIMED