convolutional neural network for detecting plant disease

Project Title

convolutional neural network for detecting plant disease

Project Area of Specialization Artificial IntelligenceProject Summary

The most recent age of convolutional neural systems (CNNs) has accomplished amazing outcomes in the field of picture characterization. This task is worried about another way to deal with the improvement of plant malady acknowledgment model, in view of leaf picture arrangement, by the utilization of profound convolutional systems. Novel method for preparing and the strategy utilized encourage a fast and simple framework usage by and by. The created model can perceive 5 unique kinds of plant infections out of solid leaves, with the capacity to recognize plant leaves from their environment. Every single fundamental advance required for executing this infection acknowledgment model are completely depicted in the expectations, beginning from social event pictures so as to make a database, surveyed by horticultural specialists.

Project Objectives

According to Environmental and Social Aspects

•Plant diseases do have many negative impacts on our daily lives that we sometimes may not be aware of.

•If our crops are not successful, we must struggle for more food. That’s why people must develop agricultural technologies to provide better and healthier crop.

•Farmers spend lot of charges on disease management, often without adequate technical support, resulting in poor disease control, pollution and harmful results.

•In addition, plant disease can devastate natural ecosystems, compounding environmental problems caused by habitat loss and poor land management.

According to Economical aspects

•By using the CNN technology for the detection of plant leaf disease is economically less rather then without use of technology and contact with experts and give them charges according to their demands and call them again when we feel that crop need a detection of disease again

•On the other hand the use of CNN technology has it’s a initial cost for a farmer and he can detect his crops whenever he want.

Project Implementation Method

Image cropping

•Image cropping reduces the amount of computation used by the GPU to minimize the foreground portion.

•Resize the image obtained from cropped image

Multilevel technique

•Multi-scale is a learning process that randomly deforms several

sizes by using the minimum and maximum sizes.

•By using this method, it is possible to prevent the

overfitting phenomena arising as a result of less learning data.

Learning using CNN

•A basic and modified structure of the LeNeT model are used the size of the input image is adjusted to 224 * 224, and a 3 * 3 stride 2, and 3 * 3 stride 1, of the convolution is performed.

•A 3*3 padded convolution operation is also performed to reduce data loss before pooling. After pooling, a 3*3 stride 1, 3*3 stride 2, and 3 *3 stride 3 convolution are executed.

Benefits of the Project

The proposed technique can take the upsides of the neural system to separate the qualities of sick parts, and subsequently to order target ailment regions. To address the issues of long preparing assembly time and too-enormous model parameters, the customary convolutional neural system was improved by joining a structure of origin module, a crush and-excitation (SE) module and a worldwide pooling layer to recognize maladies. Through the Inception structure, the element information of the convolutional layer were combined in multi-scales to improve the precision on the leaf ailment dataset.

Technical Details of Final Deliverable

•Initial dataset with 1000 number of images

•Training of Network

•Created a Model

•Train and Validation accuracy

•Train and Validation loss

Final Deliverable of the Project Software SystemCore Industry AgricultureOther Industries Food Core Technology Artificial Intelligence(AI)Other Technologies NeuroTechSustainable Development Goals Good Health and Well-Being for PeopleRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	60000
spyder + camera + laptop	Equipment	3	20000	60000