Using Deep Learning for Image Recognition: Building Deep Neural Network Models to Predict Handwritten Digits
Author: David Rodrigues | https://github.com/davidrodriguessp | https://www.linkedin.com/in/davidrodrigues/ | davidrodriguessp@hotmail.com
In this project, we created different models to predict handwritten digits. We used a dataset available in the Scikit-learn Python library called load_digits(). The dataset has 1,797 images. Initially, we created a model using the K-nearest neighbors algorithm. Then, we built different models using Artificial Neural Networks.
The conclusion was that the different models reached similar scores, but K-nearest Neighbors (KNN) with k=1 had a slightly better performance, reaching 99% accuracy.
Notebook: Deep Learning for Image Recognition.ipynb