The challenge topic: Diseases and Pests
How can technology reduce the economic impact of plant diseases and pests by improving farmers’ ability to quickly and accurately diagnose and respond to plants/trees/vines with disease or pest issues?
The following are the problems faced by farmers :
- Agriculture plays a critical role in providing food supply for growing population of the world.
- Annual global food supply loss due to plant disease is 40%, on average.
- In most of the countries, smallholder farmers generate more than 80% of the agricultural production. For them, the loss of crops has devastating consequences. 4.Sometimes, farmers can lose almost 100% of their crop due to plant disease. This makes crop disease a major threat to food security around the world. after looking at these problem comes the idea of CropStetho - An Artificial Intelligence Based Crop Diagnosis tool which take image of Leaf of crop as input and Predict the Disease with the Information on that disease and solution.
Timely and correct identification of a disease when it first appears is a critical step for efficient disease management.
An app that the farmer can have their smartphone with the following capabilities: *Identify the type of plant *Identify if the plant has the disease *If the leaves are affected by a disease, classify the disease *Provide treatment options by pulling information from the database and attaching links to online sources.
How we built it
We used Deep Learning pipeline to create such an incredible project, Steps are as follows : Adapt a pretrained model to our image classification problem AlexNet (Krizhevsky et al., 2012) + Transfer Learning Reshape the last layer: FC7 (4096 neurons) -> FC8 (neurons) Train the model with train-test ratio = (0.1 0.9) We used Pytorch to develop the model. and rest is in PPT
We used Figma as a foundation for the UI/UX interface
Challenges we ran into
First we were trying to train model on local machine which was taking a lot of time then we side by side used Google Colab's GPU's !
Accomplishments that we're proud of
That we were able to complete it in time
What we learned
A lot of working on Deep Learning Models! We also learned about the importance of treating crops and identifying the best and safest treaments as soon as a problem