Pest identification and classification

Project Title

Pest identification and classification

Project Area of Specialization Artificial IntelligenceProject Summary

Introduction:

Agricultural is the backbone of Pakistan's economy. About 68% of the population is engaged in farming directly or indirectly through production, processing and distribution of major agricultural commodities. It contributes about 24 percent of Gross Domestic Product (GDP) and accounts for half of employed labor force and is the largest source of foreign exchange earnings.

World population is growing day by day and according to World Population Prospects, the global population in 2050 will be around 9.77 billion people, which is 2 billion more than what the current population is today. To feed the growing global population, estimates suggest we’ll have to increase food production by as much as 68 percent by 2050. If we were to talk about sustainable development goals, food appears central to many of them including:

• Zero Hunger (SDG 2);
• Good Health and Wellbeing (SDG 3);
• Responsible Consumption and Production (SDG 12);
• Life on Land (SDG 15), which includes sustainable agriculture;

Summary:

There are many factors which affect the agriculture yield and production like water consumption, sunlight, fertilizers, pesticides, etc. We have studied things out and concluded that pests destroy up to 40 percent of global crops and cost $220 billion of losses.

We have targeted agriculture pests as our final year project. We are designing such a system that will identify pests available on crops and will classify those pests from our datasets using artificial intelligence based algorithms and techniques. We will use cloud services for saving and training our datasets. The pests’ identification and classification is the first step towards saving crops for farmers. But, we see our project too far in future. We aim to bring precision agriculture techniques to Pakistan Agriculture. We see that farms around the globe are adapting to PA techniques which aims tell farmers precisely what inputs are needed, where, in what amount and, when to produce more for less.

Precision Agriculture is expensive since many advanced sensors and machinery is needed in PA like Geospatial Technologies like GPS GNSS, Variable Rate Technology or VRT, smart sensors, drones, etc. We wish to start off by using easily available sensors like multi spectral sensors, IOT based sensors to measure soil health and temperature, oxygen amounts, and air temperature, water amounts, sun amounts, insects, and pesticides, raspberry pi and cloud services.

Conclusion:

We wish to contribute our part to help grow more food and achieve No Hunger SDG around the globe.

Project Objectives

Introduction:

Agriculture plays a vital role in country’s economy. We all depend upon food coming through our agriculture crops like wheat, cotton, barley, etc. The growth and production of most important field crops are affected due to attack of various pests. Early pest identification is of paramount importance in terms of productivity and reduction of the use of pesticides. Eye observation methods have been used in recent years, but they are not efficient in large crops. Plant diseases and pests are important factors determining the plant yield and production. Our project aims to develop a system which gives the farmers an easy way to detect and classify pests and also embed precision agriculture techniques for less resource usage and more production.

Objectives:

Following are the objectives of our final year project:

• We aim to make pests identification system for agriculture crops by using international open source datasets and our Pakistani clicked images.
• We also aim to make pests classification system by making up to 4 classes using our datasets.
• Another objective is to use Artificial Intelligence algorithms using Machine learning and deep learning techniques on python language to enhance our skills and automate pests system in crops.
• We are planning to use Convolutional Neural Network algorithm as well as Python based libraries like OpenCV, etc, and cloud services for the entire software system.
• We also aim to make a small prototype using IOT sensors to measure soil health and temperature, oxygen amounts, and air temperature, water amounts, sun amounts, insects, and pesticides, raspberry pi, and camera.
• We thinking that our IOT based prototype will be helping us in applying a precision agriculture concept to our farms which will be more cheaper in cost than the actual precision agriculture techniques.  
• We aim to help farmers to use less resources, in terms of human labor, soil and plant needs, more technology to produce more than what they are producing currently.
• Finally, we aim to use SW/HW in our project and play our part in achieving sustainable development goals like No Hunger.

Project Implementation Method

Introduction:

We are designing such a system that will identify pests available on crops and will classify those pests from our datasets using artificial intelligence based algorithms and techniques. We will use cloud services for saving and training our datasets. The pests’ identification and classification is the first step towards saving crops for farmers. But, we see our project too far in future. We aim to bring precision agriculture techniques using IoT to Pakistan Agriculture. We see that farms around the globe are adapting to PA techniques which aims to tell farmers precisely what inputs are needed, where, in what amount and, when to produce more for less.

Implementation:

Domain:

We in our final year project are working in artificial intelligence domain. But, on funding approval we will target Internet of Things (Iot).

Backend Implementation:

Datasets:

We have collected datasets of pests available internationally and in Pakistan. Our implementation will start from gathering requirements about common crops and pests in Pakistan and international agriculture sector. We will firstly make expectedly 4 classes out of the datasets we have. We can also take images from camera mounted in agriculture crops.

Image Segmentation and Preprocessing:

We will then apply image preprocessing and segmentation techniques on those images to get improved images.

Programming Language:

We will be using Python programming language on the development side. We will use Python based libraries like OpenCV, numpy, etc for pest identification and classification as well as embedding IoT in the system for future modules.

Algorithm:

We will then move towards model designing. Our research shows that the best artificial intelligence algorithm we can use is Convolutional Neural Network (CNN). We will develop and train CNN algorithm according to our requirements in the development phase. We will then apply that model on our datasets and check the accuracy on our dataset classes. Our trained CNN algorithm will classify the pest datasets according to our trained classes.

Cloud Service:

Our datasets will be saved, trained and tested on paid cloud services like Microsoft Azure Cloud or Amazon AWS Cloud.  

Frontend Implementation

Mobile Application:

On the front-end, we will be designing a Flutter based mobile application, that will be browsing those images from Cloud and identify and classify them and will show the result to the farmers. For the purpose of applying precision agriculture techniques, we will be taking all the information from IoT based sensors like temperature, humidity, raspberry pi, as well as drones, etc. and will display that on the mobile phone application.

Final Demonstration:

For the final demonstration, we will be also working on easily available agriculture plants in nursery to easily show the prototype in final evaluation.  

Benefits of the Project

Introduction:

The agriculture and food processing industry is among the major sectors in any country and plays an essential role in expanding the export quality of agricultural and food products. Pest attack is one of the significant problems in the agriculture sector that results in degradation of crop quality. Pests, germs, and weeds cause massive loss to crops and results in a low market for the final products. Finding new ways to gain, even small increases in efficiency can make the difference between turning them into a profit or a loss.

Our project initially is targeted on pests available on crops, their identification, and classification using AI. We also wish to embed IoT in our project to achieve some precision agriculture in Pakistani farms. Since, precision agriculture is being used globally in agriculture sector.

Benefits:

Following are the benefits of our system:

• Our project aims to standardize the traditional farming system in which the farmers manually check their crops for pests by visiting almost every day to their farms.
• Our AI based design will help us in identifying and classifying pests on crops which usually is a manual process by farmers. We through our system will automate that.
• Moreover, our design using IoT will help us in applying precision agriculture techniques which are being used globally to produce more and use less resource.
• Our system will be much cheaper comparatively to what other countries are using.
• Our designed framework will help farmers in saving time, human labor, and resources in saving their pests-affected crops and will ultimately increase their production.
• The system will be connected to the mobile applications for keeping the system handy and accessible to every farmer.
• This will promote a healthy farming experience for farmers thus increasing their production and profit rate. Also, this will help us in fighting against food security issues that is a major issue in the world.
• This will help us to take part in achieving many SDGs like No Poverty, good human health and well-being, etc.

Technical Details of Final Deliverable

Introduction:

Our project is based technologies that are being used all over the world. Like artificial intelligence, machine learning, deep learning, internet of things and cloud computing. These technologies are growing rapidly and helping a lot in providing bigger and better solutions.

Our initial idea of project is to work on identification and classification of different pests on different agricultural crops using artificial intelligence, machine learning and deep learning. This will help farmers to better take care of their crops from pests using a mobile phone application.

Our improved concept of project is to somehow apply precision agriculture techniques in a different and less costly manner to farms using IoT.

Technical Details:

Following are the technical details of our final year project:

On deep research, we found that pests, germs, and weeds cause massive loss to crops and results in a low market for the final products.

Datasets:

We are developing a system for which we have gathered and collected data sets of different crops and pests. We will apply pre-processing techniques on the raw data of images we have. Data preprocessing in Machine Learning refers to the technique of preparing (cleaning and organizing) the raw data to make it suitable for a building and training Machine Learning models.

The libraries used will be like pandas, numpy, matplotlib in Python.

Data/ Pest Identification and Classification:

For pest identification and classification we will be using python libraries like TensorFlow and algorithm like CNN. CNN is best for image data as well as it is considered to be more powerful than ANN and RNN.

Cloud:

We will be using cloud services like Microsoft Azure or Amazon AWS for saving, and training data sets as well as linking data sets to our mobile application

Mobile Application:

On successful training, testing and modeling our mobile phone application that will be Flutter based shows up the pest that is identified and classify that on the main screen to the user or farmer. Reducing human labor, resource and time and increasing overall farm production

Precision Agriculture

Sensors:

For the purpose of applying precision agriculture technique on the farms, we will be using IoT based sensors like Raspberry pi 4 Model B 2GB, SHT20 Humidity and Temperature sensor module, AR8210 Smart Sensor( Oxygen Sensor), Soil Moisture Sensor, and Biosensors to make farmers more knowledgeable about their farms so that they would not need to irrigate entire farm but only the parts which needs their care. This will reduce the entire resources, time, and human labor to increases the overall production.

Final Deliverable of the Project HW/SW integrated systemCore Industry AgricultureOther IndustriesCore Technology Artificial Intelligence(AI)Other Technologies Internet of Things (IoT)Sustainable Development Goals Zero Hunger, Good Health and Well-Being for People, Responsible Consumption and Production, Life on LandRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	61130
Raspberry pi 4 Model B 2GB	Equipment	1	24500	24500
SHT20 Humidity and Temperature sensor module	Equipment	1	1100	1100
AR8210 Smart Sensor( Oxygen Sensor)	Equipment	1	24000	24000
Soil Moisture Sensor	Equipment	1	250	250
NUCLEO-L152RE STM32 Nucleo development board	Equipment	1	6200	6200
1.5v aaa batteries	Equipment	2	90	180
Printing	Miscellaneous	4	600	2400
Photostates	Miscellaneous	4	500	2000
Bindings	Miscellaneous	2	250	500