Development of Disease Scoring System for Severity Analysis of Late Blight of Potato based on Image Processing Approach Satyendra Nath Mandal 1 *, Kunal Roy 1 , Sanket Dan 1 , Subhranil Mustafi 1 , Subra- ta Dutta 2 , Ashis Roy Barman 2 and Ashis Chakraborty 2 1 Information Technology, India 2 Plant Pathology, India Introduction Late blight, (Phytophthora infestans) is most devastating disease of potato worldwide with several instances of catastrophic extent with huge economic and humanitarian loss. Under cool night and warm day and extended wet condition, this pathogen sporulate profusely so that entire potato field can be destroyed in less than ten days. The average annual losses due to late blight was to 15% of total production in the country. Late blight results in global annual losses of potatoes that would be sufficient to feed anywhere from 80 to many hundreds of millions of people [1-10]. Global tuber yield losses were reported to be €12 billion [5]. The reasons for such losses are the manual observation of the disease in the potato field when the pathogen spread very fast so that the management become ineffective. Thus, quick and early detection of disease is an important landmark for successful management. Traditionally, plant disease severity is scored with visual inspection of plant tissue by trained experts. The inconsistent results on measurement of plant disease based on eye estimation hinders the decision support system of fungicidal management schedule of late blight of potato. One alternative method is automatic prediction of disease intensity through image-based analysis so that farmers may be able to spray effective fungicides in proper time. With the population of digital cameras and the advances in computer vision, the automated disease diagnosis models are highly demanded by precision agriculture, high-throughput plant phenotype, smart green house, and so forth. In recent years, spectral image-based disease severity scoring has gained new height, however, number of research works in this direction is meagre. Noticeably, deep learning convolutional neural networks, random forests and multilayer perceptron using band differences has been used to predict the level of infection of Phytophthora infestans on potato crops in Colombia [8-10]. Further, high-resolution portable spectral sensor was utilized to investigate the feasibility of detecting multi-diseased tomato leaves in different stages of crop growth, including early or asymptomatic stages [11-13]. Crimson Publishers Wings to the Research Research Article *Corresponding author: Satyendra Nath Mandal, Information Technology, India Submission: January 1, 2021 Published: March 04, 2021 Volume 5 - Issue 1 How to cite this article: Satyendra Nath Mandal, Kunal Roy, Sanket Dan, Subhranil Mustafi, Subrata Dutta, Ashis Roy Barman, Ashis Chakraborty. Development of Dis- ease Scoring System for Severity Analysis of Late Blight of Potato based on Image Processing Approach. Cohesive J Micro- biol Infect Dis. 5(1). CJMI. 000601. 2021. DOI: 10.31031/CJMI.2021.05.000601 Copyright@ Satyendra Nath Mandal. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License, which permits unrestricted use and redistribution provided that the original author and source are credited. ISSN: 2578-0190 1 Cohesive Journal of Microbiology & Infectious Disease Abstract The severity for late blight disease of potato is quantified based on standard disease grading scale mainly by measure with the eye, which is greatly influenced by subjective factors and results in obvious error using Henfling disease estimation scale. The plant pathologists generally graded the late blight disease severity based on eye estimation for percent infected area according to the disease grading scale of Henfling. The disease scoring system is primarily the eye estimation methods and scores are varied among different pathologists. In this paper, an attempt has been made to develop image processing- based disease estimation for late blight of potato. The late blight affected portions of leaflets, leaves and whole plants have been evaluated using image analysis system. The percentages of affected areas have been calculated and scores have been assigned based on proposed scale. The assigned scores have also been verified from several plant pathologists based on Henfling’s disease scale. The gap of disease intensity variation between pathologist’s evaluation and image processing system has been fine-tuned by modification of the scale and finally the accuracy of scores estimation based on proposed image analysis system is 85% and could be effectively exploited for late blight of potato disease estimation. Keywords: Late blight of potato; Henfling Disease estimation scale for late blight; Image based disease estimation scale; Image analysis; Plant pathology