Sugar-Beet Stress Detection using Satellite Image Time Series

2025-07-17

AI Summary

This study addresses the challenge of unsupervised identification of abiotic and biotic stress in sugar beet fields. We propose a 3D convolutional autoencoder framework incorporating acquisition-date-aligned temporal encoding to jointly model time-series spectral features from Sentinel-2 imagery. By directly learning crop growth dynamics without manual annotations, the model captures intrinsic phenological patterns. The date-specific temporal encoding significantly enhances cross-year generalizability, mitigating interannual sensor calibration and phenological variability. Downstream clustering analysis demonstrates robust discrimination between healthy and stressed fields. Experiments across multiple growing seasons confirm high consistency in stress detection, with strong practicality and transferability across years and regions. Our approach establishes a novel unsupervised paradigm for agricultural remote sensing monitoring, enabling scalable, label-free stress assessment in precision agriculture.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Machine Learning: Unsupervised & Self-Supervised Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Web Mining and Content Analysis: Models for Web evolution

Abstract

Satellite Image Time Series (SITS) data has proven effective for agricultural tasks due to its rich spectral and temporal nature. In this study, we tackle the task of stress detection in sugar-beet fields using a fully unsupervised approach. We propose a 3D convolutional autoencoder model to extract meaningful features from Sentinel-2 image sequences, combined with acquisition-date-specific temporal encodings to better capture the growth dynamics of sugar-beets. The learned representations are used in a downstream clustering task to separate stressed from healthy fields. The resulting stress detection system can be directly applied to data from different years, offering a practical and accessible tool for stress detection in sugar-beets.

Problem

• Detect stress in sugar-beet fields using satellite data
• Develop unsupervised 3D autoencoder for feature extraction
• Create tool for multi-year stress detection in agriculture

Innovation

• 3D convolutional autoencoder for feature extraction
• Temporal encodings to capture growth dynamics
• Unsupervised clustering for stress detection