Javier Garcia Molina's repositories
selfie-background-removal
In this project I built an algorithm that could automatically remove the background from a selfie. For that, I used the U-net architecture.
oxford-pets
In this project I built a model able to classify dog and cat images according to their breeds. I used the dataset from the Oxford Pets Kaggle competition, which consists of 7,349 training images and comprises 37 breeds of cats or dogs. I used Deep Learning, Computer Vision and Transfer Learning techniques.
breaking-captchas
In this project I downloaded a dataset of captcha images and went through the end-to-end process of extracting the digits from the images, labelling the ones from the training set so that I could feed them to a neural network, training a model on the digits, and using this model for labelling the whole captcha images of the test set.