Bridging the gap between contemporary remote sensing and archaeology: The S.A.D.A. platform for accessible remote-sensing image analysis

Remote sensing now supplies abundant, high-resolution imagery for archaeological prospection, yet many projects still underuse these data due to technical barriers. This paper presents S.A.D.A. (Smart Anomaly Detection Assistant), an open-source, GUI driven application that integrates common pre-processing (index calculation, histogram adjustment, PCA), multiple anomaly-detection families (Standard, RX, Isolation Forest, LOF, one-class SVM), unsupervised clustering (K-means, DBSCAN), and spatial autocorrelation tools (LISA) within a guided, human-in-the-loop workflow. We evaluate S.A.D.A. on three contrasting datasets—multispectral UAV imagery from Veio, a thermal UAV frame from Jesolo, and a pan-sharpened Pléiades-1A scene from Metaponto—illustrating how the platform can support iterative, method-agnostic exploration and export GIS-ready outputs. Results indicate that S.A.D.A. can lower the practical cost of applying established techniques across heterogeneous sensors and scales, while keeping interpretation with the analyst. The discussion considers usability (guardrails, contextual tooltips, explicit pipelines) and limits (no automatic site detection, the need for archaeological validation), and outlines development priorities, including multi-modal support and optional human-in-the-loop deep learning. We suggest that tools like S.A.D.A. may help normalize remote sensing as a routine—yet critically interpreted—component of archaeological practice, particularly where budgets and specialist capacity are limited.