#Potato Disease Classification task This is a system, which takes an uploaded image of a potato plant and using the trained model classifies it as one of the 3 classes of diseases. This project I believe has agricultural benefits and helps farmers identify early signs of disease and prevent than as early as possible. Its development is based entirely on CNNs. The resulting model has above 92 percent accuracy.

How to run:

uvicorn main:app --reload

Properties:

-- Data preprocessing and augmentation to prevent overfitting

-- CNN with multiple layers of convolution, pooling, activation following each other, and one fully connected softmax layer in the end

-- Compiled using Adam optimizer, and sparse categorical cross-entropy as loss function.

-- Model doesn't overfit: training accuracy - 93 percent, testing accuracy: 92 percent