Disease classification of winter crops by CNN

In agriculture sector, Pakistan is ranked in the top of the list. Various factors such as climate condition and various diseases effect the production of winter crops therefore their early identification is very important. The food and agricultural organizations of the world estimates that pest

Project Title

Disease classification of winter crops by CNN

Project Area of Specialization

Artificial Intelligence

Project Summary

In agriculture sector, Pakistan is ranked in the top of the list. Various factors such as climate condition and various diseases effect the production of winter crops therefore their early identification is very important. The food and agricultural organizations of the world estimates that pests and diseases lead to loss of 16%-18%.of global food production, constituting a threat to the world. Plant pathogens; also causing fungal diseases; represent relevant biotic stress factors responsible for significant crop yield losses.

Convolutional Neural Networks (CNN) have demonstrated their capabilities on the agronomical field, especially for plant visual symptoms assessment. Artificial Neural Network has been utilized for winter plant diseases classification and yield predictions. As these models grow both in the number of training images and in the number of supported crops and diseases, there exist the dichotomy of (1) generating smaller models for specific crop or, (2) to generate a unique multi-crop model in a much more complex task (especially at early disease stages) but with the benefit of the entire multiple crop image dataset variability to enrich image feature description learning. In this work we will first introduce a challenging dataset of more than one thousand images taken by cell phone in real field wild conditions. When applying existing state of the art deep neural network methods to validate the two hypothesis approaches, like BAC for smaller specific models and single multi-crop model. In this work, we will propose three different CNN architectures that incorporate contextual non-image meta-data such as crop information onto an image based Convolutional Neural Network. This combines the advantages of simultaneously learning from the entire multi-crop dataset while reducing the complexity of the disease classification tasks. The models of winter rapeseed yield produced in the work will be the basis for the construction of new forecasting tools, which may be an important element of precision agriculture and the main element of decision support systems.

Therefore, continuous plant stock controls are required to identify and classify disease symptoms in preferably early infestation stages to enable most efficient treatments. Thus convolutional neural network algorithms could provide a flexible framework that allows for the definitions of plant models that act as descriptive hierarchical feature extractor and as classifier. CNN architectures could provide 99% accuracy in classification of many diseases symptoms of winter plants. This is a time and cost intensive work.

Project Objectives

The main objectives are:

• To classify the diseases in winter crops.
• To understand the CNN and Back propagation in depth.
• To train the algorithms for better accuracy in results.
• Try to reduce the challenges occur in the previously research papers.
• To develop a mobile application which is capable of classification of normal or diseased crops, vegetables and fruits by feeding an image in it on field.

Project Implementation Method

Input: Image will be an input for our project.

Image Cropping:  This step will give the image an exact shape for the algorithm to implement.

Image to array: We have python OpenCV2 library for this on the basis of RGB ratios.

Apply CNN: In convolutional neural network we have the following models to be use:

• Conv2D: 2D Convolutional layers take a three dimensional input, typically image with three color channels. They pass a filter called convolution kernel, over the image, inspecting the small window pixels at a time, for example 3x3 or 5x5 pixels in size and moving the window until they have scanned the entire image. but we are using 3x3 kernel.

• Activation:

  • ReLU (Rectified Linear Unit):The purpose of applying the rectifier function is to increase the non   linearity in the image. The rectifier serves to break up the linearity even further in order to make up for the linearity that we might impose an image when we put it through the convolution operation.

  • Softmax: This activation is normally apply to the very last layer in the neural network instead of using ReLU or sigmoid or tanh. It is useful because it converts the output of the last layer in neural network into what is essentially a probability distribution.

• Batch Normalization: It reduces the amount by what the hidden unit values shift around.

• Max Pooling: It is sample based discretization process. The objective is to down-sample and input representation reducing its dimensionality and allowing for assumptions to be made about features contained in the sub regions binned.

• Dropout: it is a technique used to prevent the model from overfitting.

• Flatten: It is a function that converts the pooled feature map to a single column that is passed to the fully connected layer.

• Dense: It is the type of deep CNN in which each layer is connected with another layer deeper than itself. 

Plant type:  We will have the type of plant and detailed information about the input imge of plant's leave.

Prediction: After all this, we have the prediction algorithm for the image that is this a diseased or not ?

• If yes: Then tells us the type of disease or classification.
• If no: Then tells us the healthy option of plant.

Output: We have the classification of disease and the accuracy in prediction of disease by our algorithms.

Benefits of the Project

Benefits are:

• This project will give accuracy better than any other project in this field beacuase we are working on the challenges which past projects faced.
• This will be more helpful in the healthy growth of the crops in winter.
• Our project is much more easy to use by any uneducated person in farms or fields.
• This project will classify the diseased crops for the ease of the farmers.
• This will be a guider for a farmer to use pesticides on the selected crop or not.
• This project will also be markable in our GDP.
• This artificial intelligence based project will be helpful for the upcoming students.

Technical Details of Final Deliverable

• First of all it will be look like any android app which can be installed in any android mobile.
• Our project will be able to cancel the challenges faced by other researchers in this field by CNN.
• It will be free of cost for the agriculture sector.
• This project will give more accuracy than any other.
• This project will automatically highlight the diseased part of the leave by a block.
• This will predict the leaves category into healthy or non healthy and upto which rank it is healthy.
• Our android app will be easily handled by taking picture of leaves of crops on the field.
• It will take few seconds to show the output.

Final Deliverable of the Project

Software System

Core Industry

Agriculture

Other Industries

Education , Medical , Food , Health

Core Technology

Artificial Intelligence(AI)

Other Technologies

Sustainable Development Goals

Good Health and Well-Being for People, Decent Work and Economic Growth, Life on Land

Required Resources

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