mPD-APP: a mobile-enabled plant diseases diagnosis application using convolutional neural network toward the attainment of a food secure world

Abstract

The devastating effect of plant disease infestation on crop production poses a significant threat to the attainment of the United Nations' Sustainable Development Goal 2 (SDG2) of food security, especially in Sub-Saharan Africa. This has been further exacerbated by the lack of effective and accessible plant disease detection technologies. Farmers' inability to quickly and accurately diagnose plant diseases leads to crop destruction and reduced productivity. The diverse range of existing plant diseases further complicates detection for farmers without the right technologies, hindering efforts to combat food insecurity in the region. This study presents a web-based plant diagnosis application, referred to as mobile-enabled Plant Diagnosis-Application (mPD-App). First, a publicly available image dataset, containing a diverse range of plant diseases, was acquired from Kaggle for the purpose of training the detection system. The image dataset was, then, made to undergo the preprocessing stage which included processes such as image-to-array conversion, image reshaping, and data augmentation. The training phase leverages the vast computational ability of the convolutional neural network (CNN) to effectively classify image datasets. The CNN model architecture featured six convolutional layers (including the fully connected layer) with phases, such as normalization layer, rectified linear unit (RELU), max pooling layer, and dropout layer. The training process was carefully managed to prevent underfitting and overfitting of the model, ensuring accurate predictions. The mPD-App demonstrated excellent performance in diagnosing plant diseases, achieving an overall accuracy of 93.91%. The model was able to classify 14 different types of plant diseases with high precision and recall values. The ROC curve showed a promising area under the curve (AUC) value of 0.946, indicating the model's reliability in detecting diseases. The web-based mPD-App offers a valuable tool for farmers and agricultural stakeholders in Sub-Saharan Africa, to detect and diagnose plant diseases effectively and efficiently. To further improve the application's performance, ongoing efforts should focus on expanding the dataset and refining the model's architecture. Agricultural authorities and policymakers should consider promoting and integrating such technologies into existing agricultural extension services to maximize their impact and benefit the farming community.

Smartphone‐based digital phenotyping for genome‐wide association study of intramuscular fat traits in longissimus dorsi muscle of pigs

“…Pearson correlation coefficient between the digital IMF content and the IMF content measured by NIRS reached 0.74, proving the feasibility of using Pork IMF App to measure IMF content. Recently, a few methodologies based on mobile App have been successfully used to perform high‐throughput phenotyping in plants (Asani et al., 2023; Müller‐Linow et al., 2019; Röckel et al., 2022). Our research presents a case study of using a mobile device for animal phenotyping.…”

“…Mitoregulin encoded by MTLN is a conserved mitochondrial peptide that has been demonstrated to control fatty acid oxidation and regulate lipolysis in adipocytes. Accumulation of triglycerides and fat was observed in MTLN knockout mice (Averina et al, 2023;Friesen et al, 2020). NDUFAB1, a subunit of NADH dehydrogenase, is necessary for maintaining systemic glucose homeostasis.…”

Standard Methods and Innovative Techniques in Phytopathometry, the Art and Science of Plant Disease Measurement

“…Over the years plant disease detection in sub-Saharan Africa has been classically and traditionally carried out by on the spot assessment of visible plant parts due in large part to limited or non-existent access to user-friendly advanced technologies (Asani et al, 2023). Internet of things refers to collective network of connected devices and technology that facilitates communication between devices and the cloud as well as between devices themselves (Amazon, 2023).…”