SatelliteCalculator: A Multi-Task Vision Foundation Model for Quantitative Remote Sensing Inversion

2025-04-18

AI Summary

To address the weak generalization and poor transferability of quantitative remote sensing inversion models caused by multispectral characteristics and geospatial heterogeneity, this paper introduces the first vision foundation model explicitly designed for physically interpretable regression of ecological parameters—including vegetation indices, canopy structure, and carbon stocks. We propose a physics-driven synthesis method for million-scale multitask paired data and design a frozen-backbone architecture with prompt-guided cross-attention adaptation, integrating a Swin Transformer backbone, lightweight task-specific MLP decoders, and cross-attention adapters. Evaluated on the Open-Canopy benchmark, our model achieves state-of-the-art accuracy across all tasks while significantly reducing inference cost. This work demonstrates the feasibility, scalability, and physical consistency of vision foundation models for quantitative remote sensing inversion, establishing a novel paradigm for intelligent remote sensing interpretation.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Deep Neural Architectures and Foundation Models
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Quantitative remote sensing inversion plays a critical role in environmental monitoring, enabling the estimation of key ecological variables such as vegetation indices, canopy structure, and carbon stock. Although vision foundation models have achieved remarkable progress in classification and segmentation tasks, their application to physically interpretable regression remains largely unexplored. Furthermore, the multi-spectral nature and geospatial heterogeneity of remote sensing data pose significant challenges for generalization and transferability. To address these issues, we introduce SatelliteCalculator, the first vision foundation model tailored for quantitative remote sensing inversion. By leveraging physically defined index formulas, we automatically construct a large-scale dataset of over one million paired samples across eight core ecological indicators. The model integrates a frozen Swin Transformer backbone with a prompt-guided architecture, featuring cross-attentive adapters and lightweight task-specific MLP decoders. Experiments on the Open-Canopy benchmark demonstrate that SatelliteCalculator achieves competitive accuracy across all tasks while significantly reducing inference cost. Our results validate the feasibility of applying foundation models to quantitative inversion, and provide a scalable framework for task-adaptive remote sensing estimation.

Problem

• Develops a vision model for quantitative remote sensing inversion
• Addresses challenges in multi-spectral data generalization
• Enables scalable estimation of ecological variables

Innovation

• Leverages physically defined index formulas for dataset construction
• Integrates frozen Swin Transformer with prompt-guided architecture
• Uses lightweight task-specific MLP decoders for efficiency