Plant Leaf Disease Detection Application
ABOUT
Each year we lose a considerable number of crops and plantations to widespread diseases, which in turn causes a lot of monetary and labor loss and makes it difficult for farmers to sustain their livelihood. Through this project, we wanted to change their lives for the better and make it easier for them to identify these diseases in time so that they can take proper measures to safeguard their crops. We here at TreeKode Organisation want to through our project, plan to address this issue directly. We are building an interactive cross-platform application that will harness the power of deep learning and image processing to identify diseases. Our aim is to develop an algorithm that can be used to train the system on a machine to work efficiently with a relatively high success rate.
Market Demand
The potential audience for this project includes farmers and people associated with the agriculture/horticulture industry who are susceptible to plant diseases. Moreover, common people who have home garden or kitchen garden setups can also use this app to gain advantage over plant spoilage. This application could also be used by scientists and researchers as well as they could also depend on the model to detect uncommon and rare plant disorders. We basically hope to help each and everyone associated with plants.
Project Working
This project is an implementation of deep learning models in an application with the use of flutter. Using Keras API, we created a model which we trained using datasets and repeat the training to increase the accuracy of the model. The application has a material UI with a profile system where a user can create an account to login with. We first upload an image on the application which then send the image to our back-end in python to identify and detect the disease. It then begins the process of identifying if the leaf is healthy or has a disease. There is also an option to put which plant leaf the input image is of.
One assumption is that the image uploaded is that of a leaf which was chosen using the drop-box.
The result of the model is then sent back to flutter along with a cure if there is a disease.
UI Design
We have created a full-fledged UI with a profile system where there will be an upload image button which will allow the user to input an image of a leaf. There is a profile option where users can make an account to login. There are 2 app bars, one drop-down box to select the leaf on which the model has to work on.
The UI is very simplistic in the sense where there are two main buttons one to upload the image and one to start the detection of the image. The image is then uploaded to the back-end python model. The back-end then returns a solution to the disease if any, otherwise it returns an output of “Healthy leaf”
Future Work
In the future to improve on our application we will be adding more leaves to our functionality. Better accuracy and more features such as the application detecting the closest pharmacy to buy the cure for the disease are plans that we will look to complete.
GitHub Repository
https://github.com/TreeKode/Plant-Leaf-Disease-Detection-Application
