Multimodal Deep Learning for Rice Yield Prediction Using UAV-Based Multispectral Imagery and Weather Data

• DOI: 10.3390/rs15102511
• Publication Date: 2023-05-11T05:37:24Z
• Authors: Md. Suruj Mia; Ryoya Tanabe; Luthfan Nur Habibi; Naoyuki Hashimoto; Koki Homma; Masayasu Maki; Tsutomu Matsui; Takashi S. T. Tanaka
• Keywords: Robustness; Extractor; Precision Agriculture; Multispectral pattern recognition

Precise yield predictions are useful for implementing precision agriculture technologies and making better decisions in crop management. Convolutional neural networks (CNNs) have recently been used to predict yields unmanned aerial vehicle (UAV)-based remote sensing studies, but weather data not considered modeling. The aim of this study was explore the potential multimodal deep learning on rice prediction accuracy using UAV multispectral images at heading stage, along with data. effects CNN architectures, layer depths, integration methods were evaluated. Overall, model integrating UAV-based imagery had develop more precise predictions. best models those trained weekly A simple feature extractor image input might be sufficient accurately. However, spatial patterns predicted maps differed from model, although almost same. results indicated that only accuracies, also robustness within-field predictions, should assessed further studies.