Deep learning with Python

Start with Andrew Trask's two posts, Learn to Code a simple Neural Network in 11 lines of Python and Build a Neural Network using Gradient Descent Approach, which implement a network with nothing but NumPy so you can see backpropagation working. Then move to Keras with Create your first CNN, Fashion MNIST tutorial with tf.keras, the Kaggle MNIST Digit Recognizer Tutorial, and PyImageSearch's Get Started with Keras on a Custom Dataset.

Image projects make up most of the section: Break A Captcha System using CNNs by Adam Geitgey, Using Convolutional Neural Nets to Detect Facial Keypoints, Build A Facial Recognition Pipeline, Make your Own Face Recognition System, Code a Smile Classifier, Object Detection With Neural Networks, and medical imaging work such as Deep Learning and Medical Image Analysis for Malaria Detection. Two tutorials cover transfer learning, retraining a pretrained network like Inception on your own images.

For text, try Train a Language Detection AI in 20 minutes, Natural Language Processing using scikit-learn, Code a Taylor Swift Lyrics Generator, and Learn Twitter Sentiment Analysis, an 11-part series. Mask detection using PyTorch Lightning is the only PyTorch project. Many of these were written for TensorFlow 1.x or early Keras, so imports and APIs may need updating for TensorFlow 2 and Keras 3.