Crop Disease Scouting using AI and IoT Technology

Wheat is a staple food crop of Pakistan, dominating all crops in acreage and production. Wheat accounts for almost 40% of the crop area, 65 % of the food grain acreage, and 70% of the production. The diseases in wheat crop are causing problems such that the yield is not sufficient to meet the needs.

Project Title

Crop Disease Scouting using AI and IoT Technology

Project Area of Specialization

Artificial Intelligence

Project Summary

Wheat is a staple food crop of Pakistan, dominating all crops in acreage and production. Wheat accounts for almost 40% of the crop area, 65 % of the food grain acreage, and 70% of the production. The diseases in wheat crop are causing problems such that the yield is not sufficient to meet the needs. Wheat rust is one of the catastrophic fungal disease which can reduce the yield by almost 20%. To deal with this disease, a system is required which can timely detect and predict the severity level of the rust using machine learning and deep learning techniques. To address the problem, we propose an IoT & AI enabled system which captures high quality images of the infected leaf, machine learning and deep learning models will be trained to classify the rust images into three categories i.e., healthy, rust-resistant and susceptible. The learned model will be integrated into web and mobile application which will help the farmer to analyze the rust severity level. The proposed system will provide a cost-effective solution to timely detect the disease and prevent the spread, thus enhancing the crop yield.

This project is an automated multi-platform crop scouting system employing several state-of-the-art technologies like AI, IoT, CV, Edge AI, Cloud Computing for timely detection of wheat rust diseases

What is the unmet need in society that our idea will fulfill ?

The rust disease in wheat crops usually causes up to 20% reduction in crop yield. For its timely detection thousands of hours of knowledge worker time is required. Therefore, there is a dire need of an automated system that can do timely identification of rust diseases.

Who needs it ? How many would benefit ?

It is primarily needed by farmers and agriculture experts, but the need extends to a much wider spectrum, Pakistan being an agricultural country relies mainly on agriculture, therefore such a system is the need of the entire country for feeding the population

How will the solution work

A high quality camera takes the wheat leaf image which is passed through a pipeline. First segmentation is done using deep learning-based segmentation model called U2-Net. Auto cropping is then used to extract the region of interest. Deep Learning models will be trained on the cropped dataset, and this trained model will be capable of performing real time crop disease detection. Deployment will be done on AWS connected with Nvidia Jetson Nano, Android / iOS applications, and web portal.

Who are our competitors ? How is our solution different

Currently in Pakistan, there are no other fully automated cross platform solutions deployed for increasing the crop yield, making us a first-mover in the market

Project Objectives

Pakistan is an agricultural country, with 17.9% contribution to the GDP by major crop production including wheat, rice, maize, sugarcane and cotton. Agriculture is often called the backbone of the country, as it not only feeds the population, but it is also a major contributor to the GDP of the country. Wheat has specific importance in the agricultural production of Pakistan. It is the staple crop of Pakistan and feeds the entire population. Farmers in Pakistan are striving to fulfill the needs and demands of wheat and thus 80% of the farmers in Pakistan are growing wheat on their total of approximately 9 million hectares of land. Surveys conducted in Pakistan show that wheat production has faced a decline over the past 5 years and the major reasons for this decline include poor crop management, crop diseases and weather-related factors. The decline due to crop diseases is a focus of this project. Crop productions is significantly affected by crop diseases. Specifically, the wheat crop can face a decline of up to 20% due to rust disease. While, in some situations it can cause up to 50% reduction as well. This is alarming and in a country like Pakistan, farmers do not have the knowledge or resources to tackle such conditions. It is of utmost importance for saving the crop that the wheat rust be identified at an early stage before it overtakes the crop and minimizes its yield. Throughout the world, advanced techniques are being used for crop disease
detection, particularly for ease of use such solutions are being deployed to the web and mobile based applications so that it is easily accessible as well, such an application called Plantix is being used for crop disease detection, and its methodology is acknowledged by researchers in India as well, this application can detect and identify a crop disease, but it is unable to identify the various levels of rust disease. To address this important concern, a better and more detailed solution is proposed in this project which aims to detect the level of rust disease using IoT and AI based technology. For the deployment of this solution AWS will be utilized, its services of focus include AWS IoT Core and AWS Greengrass which will enable the system to be deployed on the cloud such that Machine Learning models will be trained on the cloud and these trained models will be run on the system. Moreover, the results will be displayed in the form of a web portal and the solution will be further extended to include a mobile application as well.

The main objectives are:

1. Cost effective classification & detection of wheat rust disease
2. Solution with Expandability

3. Wheat Crop Yield Enhancement

4. Predictive Model Development

5. Web Portal Development

Project Implementation Method

A high quality camera takes the wheat leaf image which is passed through a pipeline. First segmentation is done using deep learning-based segmentation model called U2-Net. Auto cropping is then used to extract the region of interest. Deep Learning models will be trained on the cropped dataset, and this trained model will be capable of performing real time crop disease detection. Deployment will be done on AWS connected with Nvidia Jetson Nano, Android / iOS applications, and web portal.

Benefits of the Project

This solution is primarily needed by farmers who face significant wheat yield reduction due to rust disease, but the need extends to a much wider spectrum. Pakistan being an agricultural country relies mainly on agriculture, therefore such a system is the need of the entire country for feeding the population and as well as to maximize a significant contributor to the country’s GDP.

The rust disease in wheat crops usually causes up to 20% reduction in crop yield. For its timely detection thousands of hours of knowledge worker time is required. Therefore, there is a dire need of an automated system that can do timely identification of rust diseases.

It is primarily needed by farmers and agriculture experts, but the need extends to a much wider spectrum, Pakistan being an agricultural country relies mainly on agriculture, therefore such a system is the need of the entire country for feeding the population

Technical Details of Final Deliverable

A system is proposed capable of automatically identifying the severity of rust disease so that a proper remedial action can be performed. The
proposed solution is an IoT and AI based crop disease scouting system that includes an IoT device with high resolution camera to capture the images of a wheat leaf for real time disease detection. A deep learning model will be utilized to classify the images into one of the three levels of wheat rust severity (healthy, rust-resistant & rust-susceptible). The captured images will be sent to the AWS IoT Core which connects the IoT device with the server on the cloud. The system will also utilize other Amazon Web Services including AWS Greengrass which will enable Machine Learning / Deep Learning models to be used on the connected IoT device and simultaneously syncing data to the cloud. The learned models will also be connected to a web portal and a mobile application for ease of use and higher availability of the system.

The overall system comprises of Data Collection, Data Preprocessing, IoT Devices, Machine Learning / Deep Learning related work, Web Development, Application Development, Integration with Amazon Web Services and testing & validation of the system.
The scope of this Final Year Project will include the following:
• Data Preprocessing
• Machine Learning / Deep Learning related work
• Web Development
• Application Development
• Integration with Amazon Web Services
• Testing & Validation of the software aspects of the system

Final Deliverable of the Project

Hardware System

Core Industry

Agriculture

Other Industries

Core Technology

Artificial Intelligence(AI)

Other Technologies

Internet of Things (IoT)

Sustainable Development Goals

Zero Hunger, Industry, Innovation and Infrastructure, Life on Land

Required Resources

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