Development of a Crop Growth Prediction Model for Vertical Farms through Multimodal-Based Fusion of Growth and Environmental Data

Abstract

The purpose of this study is to develop and evaluate a multi-modal fusion model that integrates image and environmental data collected from a vertical farm to accurately predict crop growth stages and growth rates. RGB images and key environmental parameters, including CO₂ concentration, temperature, relative humidity, and light intensity, were synchronously collected over a defined period at a vertical farm test site, and a multi-modal training dataset was constructed through preprocessing and feature extraction. Leaf area, greenness, shape indices, and proxy VI were extracted from the image data, while the temporal variations of the environmental sensor data were modeled using a long short-term memory (LSTM) network. These were then merged with convolutional neural network (CNN)–based image features to perform simultaneous growth stage classification and growth rate prediction.

Plant phenotyping relevance

画像特徴量と環境センサーデータを融合し、葉面積・緑色度・形状指標などの植物形質から生育段階と生育速度を推定する手法の開発・評価が中心である。

Code and data availability

公開本文の所在を確認できませんでした。非公開または購読が必要な可能性があります。

No evidence-backed public reproduction asset is currently recorded.