Classification, Segmentation and Visual Question Answering with Convolutional Neural Networks Why

All these projects belong to the coursework for the 2020 Polimi neural networks course. These assignments were an invaluable learning opportunity.

What

I built three convolutional neural networks for classification, segmentation, and visual question answering. The classification task involved deciding whether someone was wearing a mask or not. The segmentation task was part of a Europe Horizon 2020 challenge and required me to segment crops for precision agriculture. Finally, the visual question-answering setting provides an image and a question you need to answer given the image content.

How

All the problems have been solved with Python and Keras. PIL and OpenCV have been used for image pre-processing.