Deep Learning Sensor Fusion in Plant Water Stress Assessment: A Comprehensive Review

Abstract

Water stress is one of the major challenges to food security, causing a significant economic loss for the nation as well for growers. Accurate assessment of water stress will enhance agricultural productivity through optimization of plant water usage, maximizing plant breeding strategies, and preventing forest wildfire for better ecosystem management. Recent advancements in sensor technologies have enabled high-throughput, non-contact, and cost-efficient plant water stress assessment through intelligence system modeling. The advanced deep learning sensor fusion technique has been reported to improve the performance of the machine learning application for processing the collected sensory data. This paper extensively reviews the state-of-the-art methods for plant water stress assessment that utilized the deep learning sensor fusion approach in their application, together with future prospects and challenges of the application domain. Notably, 37 deep learning solutions fell under six main areas, namely soil moisture estimation, soil water modelling, evapotranspiration estimation, evapotranspiration forecasting, plant water status estimation and plant water stress identification. Basically, there are eight deep learning solutions compiled for the 3D-dimensional data and plant varieties challenge, including unbalanced data that occurred due to isohydric plants, and the effect of variations that occur within the same species but cultivated from different locations.

Crop phenotyping in a context of global change: What to measure and how to do it

“…These databases are mainly used in deep learning applications for plant organs and whole plant imagery (Li et al, 2020) and plant organ databases have been applied in studies of morphological traits such as leaf area and shape (Taghavi Namin et al, 2018; Dobrescu et al, 2020), as well as counting leaves (Giuffrida et al, 2018; Ubbens et al, 2018; Ghosal et al, 2019; Wang et al, 2019; Zhou et al, 2019), fruit and other organs (Bauer et al, 2019; Milella et al, 2019; Afonso et al, 2020; Gong et al, 2020; Khaki et al, 2020; Misra et al, 2020). Furthermore, deep learning has been employed to detect developmental stages like flowering (Wang et al, 2019) or to assess root architectural traits (Atkinson et al, 2019; Yasrab et al, 2019; Teramoto and Uga, 2020; Yasrab et al, 2021), water and nutrient use efficiency (Takahashi and Pradal, 2021) or even to predict yield (Bauer et al, 2019; Khaki et al, 2020) in response to unfavorable conditions, including biotic and abiotic stresses (Ghosal et al, 2018; Singh et al, 2018; Gao et al, 2020; Nagasubramanian et al, 2020; Kamarudin et al, 2021; Singh et al, 2021). Improvements to deep learning architectures can be incorporated during the design, training, and validation stages.…”

Dynamically Controlled Environment Agriculture: Integrating Machine Learning and Mechanistic and Physiological Models for Sustainable Food Cultivation

“…Water content can be measured using NIR reflectance sensors and has been used to identify water-stressed crops . Transpiration rate can be estimated through thermal measurements of leaves. − By combining dynamics equations with measurements of transpiration, irrigation, evaporation, and NIR water content, water content in the plant can achieve good observability when used in a factor graph estimator or sensor fusion. − …”

End-to-End Fusion of Hyperspectral and Chlorophyll Fluorescence Imaging to Identify Rice Stresses

“…With the ability to learn deep and representative features from big data, deep learning has been used in various fields, including plant phenotyping. Deep learning has also been used for high-throughput plant phenotyping [26,[36][37][38][39][40]. According to previous studies, shallow CNN models can work well on one-dimensional (1D) spectral data [14,41].…”