Hyperspectral imaging combined with machine learning as a tool to obtain high‐throughput plant salt‐stress phenotyping

Abstract

Summary The rapid selection of salinity‐tolerant crops to increase food production in salinized lands is important for sustainable agriculture. Recently, high‐throughput plant phenotyping technologies have been adopted that use plant morphological and physiological measurements in a non‐destructive manner to accelerate plant breeding processes. Here, a hyperspectral imaging (HSI) technique was implemented to monitor the plant phenotypes of 13 okra (Abelmoschus esculentus L.) genotypes after 2 and 7 days of salt treatment. Physiological and biochemical traits, such as fresh weight, SPAD, elemental contents and photosynthesis‐related parameters, which require laborious, time‐consuming measurements, were also investigated. Traditional laboratory‐based methods indicated the diverse performance levels of different okra genotypes in response to salinity stress. We introduced improved plant and leaf segmentation approaches to RGB images extracted from HSI imaging based on deep learning. The state‐of‐the‐art performance of the deep‐learning approach for segmentation resulted in an intersection over union score of 0.94 for plant segmentation and a symmetric best dice score of 85.4 for leaf segmentation. Moreover, deleterious effects of salinity affected the physiological and biochemical processes of okra, which resulted in substantial changes in the spectral information. Four sample predictions were constructed based on the spectral data, with correlation coefficients of 0.835, 0.704, 0.609 and 0.588 for SPAD, sodium concentration, photosynthetic rate and transpiration rate, respectively. The results confirmed the usefulness of high‐throughput phenotyping for studying plant salinity stress using a combination of HSI and deep‐learning approaches.

Utilizing spatial variability from hyperspectral imaging to assess variation in maize seedlings

“…Another way to identify subtle abiotic stress signals using spectral imaging is by correlating reflectance data with other more laborious, timeconsuming, or costly measurements correlated to the stress response. Feng et al (2019) found strong correlations between hyperspectral measurements of okra [Abelmoschus esculentus (L.) Moench] leaves with measurements linked to leaf chlorophyll content and fresh weight traditionally used to assess salt stress across crops. A common problem when analyzing spectral data of plant surfaces is taking into account uneven light scattering that occurs upon the interaction between incident light and the plant surface being captured (Makdessi et al, 2017).…”

“…Moghimi et al (2018) circumvented plant architectural differences by identifying endmembers, or single inherent values representative of a given property, indicative of all plant pixels for a given inbred genotype in a given treatment and used these to identify salt stress across wheat lines. Feng et al (2019), on the other hand, was able to develop an instance segmentation model using deep learning to segment individual okra plant leaves for further evaluation, which is a suitable approach for crops where leaves lie relatively flat horizontally with respect to the sensor.…”

Proximal Methods for Plant Stress Detection Using Optical Sensors and Machine Learning

“…Rather than collecting spectra from an entire image or an entire plant leaf, where spectra from the stressed and unaffected areas are mixed together, hyperspectral imaging can provide more sophisticated data that can isolate spectra only from the affected area and identify specific imaging patterns and characteristics. This method has become increasingly popular for plant phenotyping and stress detection in agriculture [29][30][31] and has been used to identify plant responses to both abiotic and biotic stresses, such as drought stress in maize [32] and barley [33], yellow rust [34] and powdery mildew [35] in wheat, salt stress in okra [36], and Black Sigatoka disease in banana plants [37].…”

Smart Farming Becomes Even Smarter With Deep Learning—A Bibliographical Analysis

“…Milella et al [74] proposed methods for automated grapevine phenotyping. Feng et al [75] combined machine learning with hyperspectral imaging to develop a tool for salt-stress phenotyping.…”