Unifying Heterogeneous Multi-Modal Remote Sensing Detection Via Language-Pivoted Pretraining

Yuxuan Li, Yuming Chen, Yunheng Li, Ming-Ming Cheng, Xiang Li, Jian Yang
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:66236-66254, 2026.

Abstract

Heterogeneous multi-modal remote sensing object detection aims to accurately detect objects from diverse sensors (e.g., RGB, SAR, Infrared). Existing approaches largely adopt a late alignment paradigm, in which modality alignment and task-specific optimization are entangled during downstream fine-tuning. This tight coupling complicates optimization and often results in unstable training and suboptimal generalization. To address these limitations, we propose BabelRS, a unified language-pivoted pretraining framework that explicitly decouples modality alignment from downstream task learning. BabelRS comprises two key components: Concept-Shared Instruction Aligning (CSIA) and Layerwise Visual-Semantic Annealing (LVSA). CSIA aligns each sensor modality to a shared set of linguistic concepts, using language as a semantic pivot to bridge heterogeneous visual representations. To further mitigate the granularity mismatch between high-level language representations and dense detection objectives, LVSA progressively aggregates multi-scale visual features to provide fine-grained semantic guidance. Extensive experiments demonstrate that BabelRS stabilizes training and consistently outperforms state-of-the-art methods without bells and whistles. Code: https://github.com/zcablii/SM3Det.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-li26q, title = {Unifying Heterogeneous Multi-Modal Remote Sensing Detection Via Language-Pivoted Pretraining}, author = {Li, Yuxuan and Chen, Yuming and Li, Yunheng and Cheng, Ming-Ming and Li, Xiang and Yang, Jian}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {66236--66254}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/li26q/li26q.pdf}, url = {https://proceedings.mlr.press/v306/li26q.html}, abstract = {Heterogeneous multi-modal remote sensing object detection aims to accurately detect objects from diverse sensors (e.g., RGB, SAR, Infrared). Existing approaches largely adopt a late alignment paradigm, in which modality alignment and task-specific optimization are entangled during downstream fine-tuning. This tight coupling complicates optimization and often results in unstable training and suboptimal generalization. To address these limitations, we propose BabelRS, a unified language-pivoted pretraining framework that explicitly decouples modality alignment from downstream task learning. BabelRS comprises two key components: Concept-Shared Instruction Aligning (CSIA) and Layerwise Visual-Semantic Annealing (LVSA). CSIA aligns each sensor modality to a shared set of linguistic concepts, using language as a semantic pivot to bridge heterogeneous visual representations. To further mitigate the granularity mismatch between high-level language representations and dense detection objectives, LVSA progressively aggregates multi-scale visual features to provide fine-grained semantic guidance. Extensive experiments demonstrate that BabelRS stabilizes training and consistently outperforms state-of-the-art methods without bells and whistles. Code: https://github.com/zcablii/SM3Det.} }
Endnote
%0 Conference Paper %T Unifying Heterogeneous Multi-Modal Remote Sensing Detection Via Language-Pivoted Pretraining %A Yuxuan Li %A Yuming Chen %A Yunheng Li %A Ming-Ming Cheng %A Xiang Li %A Jian Yang %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-li26q %I PMLR %P 66236--66254 %U https://proceedings.mlr.press/v306/li26q.html %V 306 %X Heterogeneous multi-modal remote sensing object detection aims to accurately detect objects from diverse sensors (e.g., RGB, SAR, Infrared). Existing approaches largely adopt a late alignment paradigm, in which modality alignment and task-specific optimization are entangled during downstream fine-tuning. This tight coupling complicates optimization and often results in unstable training and suboptimal generalization. To address these limitations, we propose BabelRS, a unified language-pivoted pretraining framework that explicitly decouples modality alignment from downstream task learning. BabelRS comprises two key components: Concept-Shared Instruction Aligning (CSIA) and Layerwise Visual-Semantic Annealing (LVSA). CSIA aligns each sensor modality to a shared set of linguistic concepts, using language as a semantic pivot to bridge heterogeneous visual representations. To further mitigate the granularity mismatch between high-level language representations and dense detection objectives, LVSA progressively aggregates multi-scale visual features to provide fine-grained semantic guidance. Extensive experiments demonstrate that BabelRS stabilizes training and consistently outperforms state-of-the-art methods without bells and whistles. Code: https://github.com/zcablii/SM3Det.
APA
Li, Y., Chen, Y., Li, Y., Cheng, M., Li, X. & Yang, J.. (2026). Unifying Heterogeneous Multi-Modal Remote Sensing Detection Via Language-Pivoted Pretraining. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:66236-66254 Available from https://proceedings.mlr.press/v306/li26q.html.

Related Material