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

TL;DR

This paper adapts the Segment Anything Model for satellite imagery to automatically delineate agricultural fields, introducing a fine-tuning strategy and a new regional dataset to improve segmentation accuracy and generalization.

Contribution

It presents a fine-tuning approach for SAM on satellite imagery and introduces ERAS, a new regional dataset for agricultural field delineation.

Findings

SAM-based pipeline achieves robust segmentation accuracy.

The regional dataset ERAS enhances generalization capabilities.

Fine-tuning improves performance on satellite imagery.

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.