High-Throughput Analysis of Leaf Chlorophyll Content in Aquaponically Grown Lettuce Using Hyperspectral Reflectance and RGB Images.

Chlorophyll content reflects plants' photosynthetic capacity, growth stage, and nitrogen status and is, therefore, of significant importance in precision agriculture. This study aims to develop a spectral and color vegetation indices-based model to estimate the chlorophyll content in aquaponically grown lettuce. A completely open-source automated machine learning (AutoML) framework (EvalML) was employed to develop the prediction models. The performance of AutoML along with four other standard machine learning models (back-propagation neural network (BPNN), partial least squares regression (PLSR), random forest (RF), and support vector machine (SVM) was compared. The most sensitive spectral (SVIs) and color vegetation indices (CVIs) for chlorophyll content were extracted and evaluated as reliable estimators of chlorophyll content. Using an ASD FieldSpec 4 Hi-Res spectroradiometer and a portable red, green, and blue (RGB) camera, 3600 hyperspectral reflectance measurements and 800 RGB images were acquired from lettuce grown across a gradient of nutrient levels. Ground measurements of leaf chlorophyll were acquired using an SPAD-502 m calibrated via laboratory chemical analyses. The results revealed a strong relationship between chlorophyll content and SPAD-502 readings, with an R2 of 0.95 and a correlation coefficient (r) of 0.975. The developed AutoML models outperformed all traditional models, yielding the highest values of the coefficient of determination in prediction (Rp2) for all vegetation indices (VIs). The combination of SVIs and CVIs achieved the best prediction accuracy with the highest Rp2 values ranging from 0.89 to 0.98, respectively. This study demonstrated the feasibility of spectral and color vegetation indices as estimators of chlorophyll content. Furthermore, the developed AutoML models can be integrated into embedded devices to control nutrient cycles in aquaponics systems.

Detection Method of Fungal Spores Based on Fingerprint Characteristics of Diffraction-Polarization Images.

The most significant aspect of promoting greenhouse productivity is the timely monitoring of disease spores and applying proactive control measures. This paper introduces a method to classify spores of airborne disease in greenhouse crops by using fingerprint characteristics of diffraction-polarized images and machine learning. Initially, a diffraction-polarization imaging system was established, and the diffraction fingerprint images of disease spores were taken in polarization directions of 0°, 45°, 90° and 135°. Subsequently, the diffraction-polarization images were processed, wherein the fingerprint features of the spore diffraction-polarization images were extracted. Finally, a support vector machine (SVM) classification algorithm was used to classify the disease spores. The study's results indicate that the diffraction-polarization imaging system can capture images of disease spores. Different spores all have their own unique diffraction-polarization fingerprint characteristics. The identification rates of tomato gray mold spores, cucumber downy mold spores and cucumber powdery mildew spores were 96.02%, 94.94% and 96.57%, respectively. The average identification rate of spores was 95.85%. This study can provide a research basis for the identification and classification of disease spores.

Erratum to "Phenotypic Analysis of Diseased Plant Leaves Using Supervised and Weakly Supervised Deep Learning".

[This corrects the article DOI: 10.34133/plantphenomics.0022.].

Wheat Kernel Variety Identification Based on a Large Near-Infrared Spectral Dataset and a Novel Deep Learning-Based Feature Selection Method.

Near-infrared (NIR) hyperspectroscopy becomes an emerging nondestructive sensing technology for inspection of crop seeds. A large spectral dataset of more than 140,000 wheat kernels in 30 varieties was prepared for classification. Feature selection is a critical segment in large spectral data analysis. A novel convolutional neural network-based feature selector (CNN-FS) was proposed to screen out deeply target-related spectral channels. A convolutional neural network with attention (CNN-ATT) framework was designed for one-dimension data classification. Popular machine learning models including support vector machine (SVM) and partial least square discrimination analysis were used as the benchmark classifiers. Features selected by conventional feature selection algorithms were considered for comparison. Results showed that the designed CNN-ATT produced a higher performance than the compared classifier. The proposed CNN-FS found a subset of features, which made a better representation of raw dataset than conventional selectors did. The CNN-ATT achieved an accuracy of 93.01% using the full spectra and keep its high precision (90.20%) by training on the 60-channel features obtained via the CNN-FS method. The proposed methods have great potential for handling the analyzing tasks on other large spectral datasets. The proposed feature selection structure can be extended to design other new model-based selectors. The combination of NIR hyperspectroscopic technology and the proposed models has great potential for automatic nondestructive classification of single wheat kernels.