A project made for my ML university class in which I implement a CNN (Convolutional Neural Network) and some scikitlearn algorithms (MNB, KNN, SVM, Gradient Boosting, MLP Classifier) to train on 30000 labeled monochrome images, run validation on 5000 labeled monochrome images and classify another 5000 images with labels between 0 and 8. The algorithms were used for a Kaggle Competiton and the best one was the CNN one with an accuracy of 91%.
Kaggle competition link: https://www.kaggle.com/c/ai-unibuc-24-22-2021 .