Learning Project: Number Detector 🔢
This repository is for the educational script that I developed for a number detector. It uses the MNIST dataset and trains a CNN using keras/tensorflow
The model also has an app.py, so that students can see that their script is actually running in real time. They can draw a number using the OpenCV canvas, and then submit it to the model for a prediction. The model has a 98 percent accuracy, and students can manipulate the script by changing the epochs, the slide, the number of layers and sizes of the layers
This was an educational model for students learning about CV. This was to demonstrate the pipeline with keras, as the repository below does everything with pure maths and demonstrates everything contained in a CNN as class structure
Prerequisites
this model was created after the students developed the CNN from scratch. Attached below is the repository to that project