Segment Anything for Satellite Imagery: A Strong Baseline and a Regional Dataset for Automatic Field Delineation

2025-06-19

AI Summary

Addressing the challenges of automated agricultural field boundary extraction from high-resolution satellite imagery and its reliance on costly ground surveys, this paper introduces the first adaptation of the Segment Anything Model (SAM) to the farmland delineation task. We propose a remote-sensing-specific fine-tuning strategy and a region-aware data collection methodology to establish a strong baseline model, rigorously evaluating its cross-regional generalization capability. To fill the geographic coverage gap in existing datasets, we release ERAS—the first open-source dataset specifically targeting emerging agricultural regions. Experiments demonstrate that our method significantly improves boundary localization accuracy across multi-source satellite imagery, achieving an average 8.2% IoU gain over baseline models. Both the ERAS dataset and implementation code are publicly available, establishing a new benchmark for semantic segmentation research and applications in agricultural remote sensing.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Machine Learning: Large Multimodal Models (LMMs)
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Accurate mapping of agricultural field boundaries is essential for the efficient operation of agriculture. Automatic extraction from high-resolution satellite imagery, supported by computer vision techniques, can avoid costly ground surveys. In this paper, we present a pipeline for field delineation based on the Segment Anything Model (SAM), introducing a fine-tuning strategy to adapt SAM to this task. In addition to using published datasets, we describe a method for acquiring a complementary regional dataset that covers areas beyond current sources. Extensive experiments assess segmentation accuracy and evaluate the generalization capabilities. Our approach provides a robust baseline for automated field delineation. The new regional dataset, known as ERAS, is now publicly available.

Problem

• Adapting SAM for agricultural field delineation
• Creating a regional dataset for broader coverage
• Evaluating segmentation accuracy and generalization

Innovation

• Fine-tuning SAM for field delineation
• Creating complementary regional dataset ERAS
• Assessing segmentation accuracy and generalization