A review of remote sensing-based crop yield estimation: machine learning techniques and environmental, algorithmic,...

Area of Science:

• Agricultural Science
• Remote Sensing Technology
• Data Science in Agriculture

Background:

• Agricultural technologies increasingly focus on enhancing output predictability and reliability through innovations.
• Remote Sensing (RS) offers precise, scalable solutions for large-scale agricultural monitoring and analysis.
• Crop yield estimation is crucial for food security and agricultural management.

Purpose of the Study:

• To systematically classify Remote Sensing-based methodologies for crop and plant yield estimation.
• To analyze the performance of different approaches, particularly Machine Learning (ML) and Deep Learning (DL) models.
• To discuss the limitations and future research directions in RS-based yield estimation.

Main Methods:

• Classification of RS methodologies into Sensor-Based, Platform-Based, Analytical/Modeling-based, and ML-driven models.
• Systematic review and analysis of existing studies on RS for crop yield estimation.
• Evaluation of Deep Learning architectures against key performance metrics.

Main Results:

• Deep Learning (DL) models consistently demonstrate superior performance in accuracy, precision, recall, and F1-score for yield estimation.
• DL's advantage lies in its ability to learn complex patterns, handle large datasets, and reduce manual feature engineering.
• Identified limitations span environmental, algorithmic, hardware/operational, and Wireless Sensor Network (WSN) factors.

Conclusions:

• A structured classification framework aids in understanding and addressing challenges in RS-based crop yield estimation.
• Deep Learning models represent the state-of-the-art for accurate agricultural yield prediction.
• Future research should focus on overcoming identified limitations to further enhance RS applications in agriculture.