Recognition of handwritten characters with convolutional neural network using tensorflow 2.3. Used datasets: A-Z Handwritten Alphabets, MNIST. My motivation to make this project was to acquire some skills in deploying machine learning models so that anyone can use it.
Live demo available here.
To use it just draw any digit or latin alphabet letter and press Predict button.
This site is build using flask + bootstrap. Drawing part is made using HTML5 Canvas and JavaScript.
This project use convolutional neural network written using Tensorflow to make predictions. It was trained on combination of MNIST and A-Z Handwritten Alphabets. For training I used simple augmentation (rotation and scaling). Training notebook is available here.