diagnosis of early blight disease in tomato plant based on visible/near-infrared spectroscopy and principal components analysis- artificial neural network prior to visual disease symptoms

Early diagnosis of plant diseases before the occurrence of symptoms can reduce the loss of the yield and increase the quality of agricultural crops. it also reduces the consumption of pesticides, environmental risks, and the cost of production. for this reason, the objectives of the present study were nondestructive diagnosis of early blight of tomato plant and discrimination of the most important agents of early blight (a. solani and a. alternate) in the primary stages of incidence of the disease before appearing visual symptoms using visnir spectroscopy (400900 nm). the spectral data were acquired from the leaves of the plants infected with a. solani and a. alternate, 48 hours, 72 hours, 96 hours, and 120 hours after inoculation. to develop the recognition model based on the spectral data, principal components analysis (pca) coupled with artificial neural network (ann) was used. the results showed that the pcaann model could diagnose the infected plants and pathogen species with accuracy of 93100% for test set samples. in 96 hours after inoculation, in addition to the simpler model (8 pcs and 3 neurons in hidden layer), accuracy of 100% was obtained. at all times after inoculation, there was no error in diagnosis of the plants infected with a. solani that is more pathogenic and aggressive than other species, from healthy plants. early blight in tomato plant and the type of pathogen before visual symptoms, without any plant sample preparation, could be diagnosed nondestructively (with accuracy of 93100%) using visnir (400900 nm) spectroscopy coupled with pcaann. it was concluded that this technology could be used for rapid, lowcost, and early diagnosis of this disease in tomato plant instead of timeconsuming, expensive, and destructive laboratory methods.

Diagnosis of Early Blight Disease in Tomato Plant based on Visible/Near-Infrared Spectroscopy and Principal Components Analysis- Artificial Neural Network Prior to Visual Disease Symptoms

Early diagnosis of plant diseases before the occurrence of symptoms can reduce the loss of the yield and increase the quality of agricultural crops. It also reduces the consumption of pesticides, environmental risks, and the cost of production. For this reason, the objectives of the present study were nondestructive diagnosis of early blight of tomato plant and discrimination of the most important agents of early blight (A. solani and A. alternate) in the primary stages of incidence of the disease before appearing visual symptoms using VisNIR spectroscopy (400900 nm). The spectral data were acquired from the leaves of the plants infected with A. solani and A. alternate, 48 hours, 72 hours, 96 hours, and 120 hours after inoculation. To develop the recognition model based on the spectral data, principal components analysis (PCA) coupled with artificial neural network (ANN) was used. The results showed that the PCAANN model could diagnose the infected plants and pathogen species with accuracy of 93100% for test set samples. In 96 hours after inoculation, in addition to the simpler model (8 PCs and 3 neurons in hidden layer), accuracy of 100% was obtained. At all times after inoculation, there was no error in diagnosis of the plants infected with A. solani that is more pathogenic and aggressive than other species, from healthy plants. Early blight in tomato plant and the type of pathogen before visual symptoms, without any plant sample preparation, could be diagnosed nondestructively (with accuracy of 93100%) using VisNIR (400900 nm) spectroscopy coupled with PCAANN. It was concluded that this technology could be used for rapid, lowcost, and early diagnosis of this disease in tomato plant instead of timeconsuming, expensive, and destructive laboratory methods.