SZAB CROP Deep convolutional neural networks for tomato leaf disease prediction

Deep convolutional neural networks (CNNs) have increasingly become prominent for image classification in numerous domains. Over the span of previous few years, high performance capability in terms of accuracy, computer vision, computation and deep learning methodologies have become renounced in plan

Project Title

SZAB CROP Deep convolutional neural networks for tomato leaf disease prediction

Project Area of Specialization

Artificial Intelligence

Project Summary

Deep convolutional neural networks (CNNs) have increasingly become prominent for image classification in numerous domains. Over the span of previous few years, high performance capability in terms of accuracy, computer vision, computation and deep learning methodologies have become renounced in plant disease classification. The plant diseases affect the economic significance of the plants and its products, along with abating their economical prominence. The tomato plant, specifically its leaves are highly affected by numerous diseases. This project aims to solve this real world problem by developing a software system which will be effective and appropriate method for diagnosis of tomato leaf diseases and its respective symptoms. In this research based project traditional CNN models, which use features extracted from CNN and employ traditional classification method are used and more robust deep neural network architecture based on transfer learning is also used. The system will be cost effective and help ensure timely diagnosis of the disease.

Project Objectives

The Project is aimed at creation of a software system which takes input inform of an image of the leaf of tomato plant, processes it and displays the output whether crop is diseased or normal. If diseased, then which disease and recommended steps would be suggested so the growers could rectify the issue.

The disease of crop is a natural phenomenon and it cannot be stopped however corrective measures if taken in due time could reduce the damage to the crop thereby livelihood of people and in general to the economy of Pakistan.

Project Implementation Method

The software system will be developed in python. The software system will have an interactive graphical user interface where the user could select the model he or she wishes to use from four of AI models and select the image from the directory or capture in real time using the webcam. The input will be processed by the model which is developed using convolution neural networks methodologies. The model will be trained on a dataset of diseased and non diseased images of leaves of tomato from the dataset. The feature extraction in the model would be done with the help of convulational layers followed by sub sampling which could be performed by maxpooling, which shall extract the features from feature map and generate the feature map in accordance with the maximum value. Furthermore, multi class classification would be done. The models will be saved and integrated into the software system.

Benefits of the Project

The plant diseases including fungal diseases are the source of reducing the yields in terms of quantity and quality of overall agricultural production. The diseases can impact many of the parts of the plants namely; fruits, vegetables, stem and leaves. The report published by the Food and Agriculture Organization (FAO) estimates that the population of world shall increase to more than 9 billion by the year 2050, this shall require a 70% growth in the production of food to ensure a steady supply. The global impact of this problem of the plant diseases and its impact on the future of mankind was the driving motivator towards pursuing our research. 

Technical Details of Final Deliverable

The final deliverable will be a software application. It shall have an interactive Graphical User Interface on the front end. The user will be able to select one of four models namely; Alexnet, VGG16, ResNet50 and ResNet152V2. Then either select single image from directory or select a whole directory or capture using live webcam session. After image has been loaded it will be shown in the software then after pressing the analyze button the image will be evaluated and the results will be displayed. The results shall include; diagnosis of name of tomato leaf disease, symptoms associated with it and the suggested treatment. 

On the backend four AI trained models will be deployed which will take image as an input analyze it with its pretrained weights and deep layers (convolutional, max pooling, activation, dense layers) and the resultant prediction be supplied to front end where result is displayed to the user. 

Each model has different architecture, which means the number of layers, their order, the neurons within them and the synapses or connections between them vary. 

Final Deliverable of the Project

Software System

Core Industry

Agriculture

Other Industries

Education , IT , Food

Core Technology

Artificial Intelligence(AI)

Other Technologies

Sustainable Development Goals

Zero Hunger, Good Health and Well-Being for People

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com