International Journal of Computer Applications (0975 – 8887) Volume 183 – No. 18, July 2021 1 Maize Lethal Necrosis Disease Detection for Maize Crop Real-Time Prediction Yield Modeling through Colour Pixel Feature Waheed Sanya Department of Computer science and IT The State University of Zanzibar Zanzibar Hashim Chande School of Agriculture The State University of Zanzibar Zanzibar Haji Ali Haji Department of Computer science and IT The State University of Zanzibar Zanzibar ABSTRACT Maize Lethal Necrosis Disease (MLND) - is one of the diseases threatening maize production in a large area of East Africa. The disease is initiated by Maize Chlorotic Mottle Virus (MCMV) in blend with viruses of genus Potyvirus, commonly Sugarcane Mosaic Virus (SCMV). The simultaneous infection is the results in intensive to complete yield loss. Inability to predict disease parameters affecting maize crop yield has been a major drawback for the effectiveness and perfection of the existing manual maize crop yield prediction system and procedure in East Africa. Presently, human visual analysis is the furthermost commonly used method for detecting diseases. Due to this method, many errors were observed as the diagnosis is mainly based on the familiarity of the farmers. It consumes time to identify crop diseases founded on visually noticeable characteristics. This research sought to propose a real-time prediction system for maize crop yield using image-based mobile for detecting crop disease affecting crop production using SVM algorithms. In the proposed model, images of maize leaves from mobile were extracted their colour features and identify the Maize Lethal Necrosis Disease (MLND). The presence of SVM due to its fast processing speed as well as accuracy of its output these algorithms requires input training and test data for the model. This prediction model will be integrated into the mobile device for farmers to use. It determines the crop leaf area index that is helpful in predicted yields and its corresponding approximate. Evaluation is also conducted against the proposed model to measure the accuracy of a real- time prediction system in producing the results of crop maize disease. The results show for the SVM, the correlation(R) between estimated Leaf Area for maize and Leaf Area affected in Tunguu area was reported as 0.6959 and 06.099 respectively. So SVM classifiers offer good accuracy as well as perform faster prediction related to naïve Bayes algorithm. Since a combination of Real-time system-based farmers mobile application images collection and Leaf Area index has never been used in East Africa so far for researcher knowledge. General Terms In this paper, the real-time prediction yield modeling througha color pixel is considered a general term. During this research, present as well as past studies of different method and technique of real-time prediction in colour features is considered to improve the algorithm in image processing. Keywords Maize Lethal Necrosis Disease (MLND), ANN, SVM, LAI, Mobile application, Real-time systems. 1. INTRODUCTION Cropyield is the measure of crop manufactured per area of land. It is an important metric toapprehend because it helps to understand food security. Achieving maximum crop yield at minimum cost with a healthy ecosystem is one of the main goals of agricultural production. Early detection and management of problems associated with crop yield restrictions can help increase crop yield and subsequent profit; hence, yield estimation is important to numerous crop management and business decisions [1]. The adoption of moreefficient farming practices and technologies can play a crucial role in improving agriculture productivity, household income, food security, and poverty reduction. This is mostly true in Sub-Saharan Africa (SSA), where the majority of the rural households are Smallholder farmers who depend on agriculture for their livelihood [2]. Traditional crop yield prediction is performed by considering and collecting farmers’ annual production values and experiences. However, crop diseases, climate, and weather change phenomena may have a considerable impact on crop yield, which leads to uncertainty in predictions [3]. Apart from climate and weather change, a disease acts as one of the contributing factors in poor production and productivity of the most important staple foods including maize [4]. Due to this fact, preventive measures are needed for early detection of diseases. The historical background of the disease in EastAfrica. The disease was primary reported in Kenya; in September 2011, even though its extent at that point recommended that the disease has existed for some time. Agreeing to the Kenyan Ministry of Agriculture, two percent of the maize harvest was affected in 2012. In August 2012, symptoms similar to MLND have also spread rapidly into Tanzania, [5]. Maize is one of the most important staple food and income-generating crop in East Africa as well as perilous for food security in Africa, supporting over 850 million people in sub-Saharan Africa[6]. It has especially significant in smallholder farming systems. However, its yield is low due to several foliar diseases. In Tanzania, maize crops contributed to 50% of the rural household incomes [7]. Currently, the crop is under threat of viruses, specifically those causing lethal necrosis disease (MLND) [8]. In Tanzania for example, Maize crop has bigger potential because it is one of the most dominant crops and the focus of a large government agricultural program [9]. The scope of this study is to address the problem in Tanzania.