Cotton Crop Disease Detection and Prevention Suggestion System

Project Title

Cotton Crop Disease Detection and Prevention Suggestion System

Project Area of Specialization Artificial IntelligenceProject Summary

Agriculture is considerably the most important necessity of mankind. Mainly countries in South East Asia like Pakistan depend upon the income that comes through this medium. The lack of implementation of technology in the agricultural field has pushed us backward. Since the income of every nation somewhat depends on agriculture, the demand to protect it is very high.
Cotton is used worldwide making clothes, medical aids or home essentials. The cotton crop fields are being damaged due to genetic diseases and exposure to various pesticides. Thus, there is a need to develop a model to identify the type of disease using modern machine learning approaches.
Our idea for the final year project is to create a system for the classification of diseases in Cotton crops and their prevention suggestions. The technique used to classify the cotton crop disease is based on using cotton crop images via deep learning-based models. The target of the system is to identify the types of diseases affecting the yield of cotton crops. This system will bring ease to the farmers (Especially amateurs) to treat the cotton crop in initial stages of disease to keep the cotton crops safe.
Our proposed model will be based on multiclassification of the cotton crop diseases by examining its cotton leaf images. Some of the diseases that will be focused are as follows: Bacterial Blight, Fusarium Wilt, and Leaf Curl Virus. After classifying the type of disease, our system will provide basic recommendations for usage of pesticides including expert opinion.

Project Objectives

The problem is that a platform does not exist in our country and which can provide service to farmers as per their need. Crops are subject to a variety of diseases that can be cured if the disease is identified on time. Traditionally diseases are identified by visually inspecting certain features like texture, color and shape of leaves. It is not an effective way for large farms or areas of Crop.
The Cotton crop is sensitive to unexpected rains, high temperature at flowering stage and pest attacks etc. So, most farmers hire professional agriculturists to diagnose their crops for diseases in large farms. However it also takes a lot of time to examine the full land crop and is very much expensive and inefficient. The more time treatment takes, the more disease will affect the crop.

The project aims to achieve the following objectives

1. Contribution to Agriculture Sector
   1. To implement Deep Learning technology in helping farmers to prevent their crops from diseases.
2.  Ease of Access.
   1. To facilitate the users (mainly famers) with the rapid detection of diseases in the crops.
3. Ecommerce Availability
   1. To facilitate users with an ecommerce portal of pesticides so they can have all the range of chemicals within the reach of a smartphone.

Project Implementation Method

Our Project is based on three tier architecture. It's a software engineering concept used for the design and implementation of software systems using client/server architecture divided into three tiers. This Separates design and implementation complexity.

 Three Layers:

• Presentation tier (user interface)
• Business logic tier (Application server)
• Data storage tier (Database server)

Our Project is divided into 3 Main Modules:

1. A Website
2. Mobile Application
3. Deep Learning Model

The Website and Mobile application will be built using three tier architecture and the Deep Learning Model will be built using CNN (Convolutional Neural Network).

Website and Mobile Application has Following Modules

Deep Learning Model has Following Modules

Follwing Tools and Frameworks will be used

1. Frontend
   1. React (Web)
   2. ReactNative(Mobile Application)
2. Backend
   1. NodeJS
3. Database Server
   1. MongoDB

        Website and Mobile Application has Following Modules

                       

       Deep Learning Model has Following Modules

          

Benefits of the Project

Traditionally all the work is done manually. Now we are going to propose a solution. Which is based on AI, Web and Android technologies that act as One Point Solution for farmers. 

In this case, mobile and web application will provide disease detection and prevention suggestions. They can get the solution (disease identification) by uploading/taking the picture of leaves taken by mobile phone or any other device. After disease detection, our platform will generate a brief report based on findings of the Deep Learning model, see Figure 1.1. The report will contain:

1. the disease name
2. the description
3. cause of disease
4. pesticides for the treatment
5. pesticides cost
6. Preventive measures.

Figure 1.1 Cotton Disease Classification report

The users will have to fill in the details required such as: size of land, soil condition, humidity, etc. to suggest the usage of pesticides and expert opinions. One more feature will be added to the report if a user is going to upload at least 25 images of land taken with a distance of +/- 20 meters that should be GPS enabled. Figure 2.1 illustrates how a Heat Map Diagram of land will be created which differentiates the normal area of the crop from the affected area using different colors.

Plus an Ecommerce portal will be added to the system. The user can buy pesticides from certified brands at best available prices. Also the videos and text description will be available on the portal which will provide the step by step guidelines about the proper way of applying the chemicals.

Figure 2.1 HeatMap Diagram

Figure 1.1 Cotton Disease Classification report

Figure 2.1 HeatMap Diagram

Technical Details of Final Deliverable

 Final Deliverable will be centralized mobile application and website. The mobile and Web applications have  Deep Learning Model integrated which will provide Cotton Crop Disease Detection and Prevention Suggestions. The  Ecommerce portal will be for pesticides through which customers can buy products accordingly. The system will play the role of facilitating users (mainly farmers) to detect diseases in cotton crops and provide prevention suggestions of the diseases. Moreover, keeping the demographic facts in view, the proposed platform will be able to have two languages Urdu and English which will create an inclusive online market environment and widen the customer net.

The Operating environment of Mobile Application and Website is as below.

• Centralized database
• Client/server system
• Operating system: Android Smartphone/Any Browser
• Database: MongoDB
• Platforms: Node.js, React.js and Flutter or React Native

The system has following constraints:

1. The Internet is a must for mobile application and website since it fetches data from the database over the internet.
2. The mobile application is for Android smartphones having Android OS 5.0 or above.
3. The smartphone must have a camera since the application will open the camera for taking a picture of the crop as the input for the disease detection.

Following are the assumptions for the system

• The user must have the internet available.
• The user must be able to read in Urdu/English.

Functional Hierarchy of System

Final Deliverable of the Project Software SystemCore Industry ITOther Industries Agriculture Core Technology Artificial Intelligence(AI)Other Technologies Cloud InfrastructureSustainable Development Goals Decent Work and Economic Growth, Industry, Innovation and InfrastructureRequired Resources

Deep Learning Model has Following Modules