Disease Identification and Severity Level Estimation on Plant

Computer Engineering

Around the world, tomato plants are the most nutrient-dense crop grown. Additionally, it significantly affects the expansion of the agricultural economy in terms of exports and cultivation. In addition to their protein content, plants also have pharmacological qualities that protect against ailments like "high blood pressure, hepatitis, gingival bleeding," etc. They are used extensively these days, which has led to a growth in the market for plants worldwide. According to statistics, over 80% of plants are produced by small farmers; as a result, insects and diseases cause more than 50% of the economic losses. Research on agricultural disease detection is particularly crucial because pathogens and insect pests are the main factors influencing plant development. Plant disease control is a challenging procedure that necessitates ongoing attention throughout the growing season and accounts for a significant portion of total production. Early detection could reduce the likelihood of yield loss, lessen the severity of chemical pollutants, and drastically reduce treatment costs. Due to the large number of plants in commercial greenhouses and the limited number of early-stage disease indicators, current disease diagnosis techniques are limited in the amount of time needed for trained personnel to physically identify and assess the pathogens. Typically, outbreak exploration is confined to sporadic cycles or limited sampling due to the expense and complexity of illness detection. The automatic detection procedures have been studied using spectroscopy, molecular processing, and analysis of volatile organic molecules. However, their implementation on a real-time operating scale is inefficient and expensive. Experiments using distinguishable data captured by conventional RGB cameras have shown the potential of machine learning techniques to detect the presence of plant diseases using deep convolutional neural network models.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.