Anomaly Detection with a Neural Network (all-CAE)

For my bachelor’s thesis in Computer and Systems Engineering, I developed a specialized neural network model known as an all-convolutional autoencoder (all-CAE) to differentiate between healthy berries and those that are infected or damaged. The all-CAE leverages the power of convolutional layers to accurately identify and classify the subtle differences in berry conditions, showcasing the potential of deep learning in agricultural applications. I used Python with Google Colab to leverage GPU acceleration.

Both my thesis and the code can be found here.