Detection Disease Tomato Using Cnn

EasyChair Preprint 11622

Abstract

Tomato production has been rising over recent years. Apart from nutritional values, tomato production also plays a significant role in terms of creating job opportunities for a lot of people. Even though the most suitable season for producing tomato is winter, tomato cultivation has been proved profitable throughout the year. However, the production of tomatoes gets hindered due to various diseases of tomato leaves. The goal of our research is to create a model that can predict these diseases earlier with the intent that suitable actions can be taken to mitigate the situation. This work presents a deep CNN model with over 22500 images divided into 3 parts: Training, validation, and testing. The dataset was prepared using necessary normalization, augmentation, and label encoding techniques. Total 10 classes (9 diseases and 1 healthy class) were taken into consideration. 75 instances were separated from each class for testing purposes. 80% of the remaining data was used for training and 20% was used for validation. The proposed Deep CNN is trained with various dropout values and a suitable dropout value is identified to regularize the model. The model was eventually able to achieve a high accuracy of 98.77% for the considered dataset. The outcomes of the work shows the effectiveness along with feasibility of our suggested model .

Keyphrases: CNN, Meetric, deep learning, loss

@booklet{EasyChair:11622,
  author    = {Youssef Laatiri and Mahjoub Mohamed and Karim Kalti},
  title     = {Detection Disease Tomato Using Cnn},
  howpublished = {EasyChair Preprint 11622},
  year      = {EasyChair, 2023}}