This is a basic template for training a model with Tensorflow
You can either clone this repo or save a copy of the jupyter notebook with this colab link. All code is located in HackMIT_Hackpack_Facial_Detection_with_Webcam.ipynb
The notebook HackMIT_Hackpack_Facial_Detection_with_Webcam.ipynb is a classic example of a computer vision project that trains a CNN to detect the presence of faces. The notebook walks through a step-by-step explaination of the problem, training data, CNN model architecture, and training and testing process. Within the notebook, you will learn how to visualize your data, create a Tensorflow model, use your webcam to test out your facial recognition model, and use the .keras extension to load and save your model. The notebook challenges you to think critically about the impact your machine learning project may have on its users, notice its bias, and improve upon the baseline model.
This hackpack is adapted from curriculum develop by Harvard and MIT Momentum AI; the code and method is adapted from Amini et al. AIES 2019, and MIT's Introduction to Deep Learning Class.
For more resources to get started with a machine learning project, we recommend some sources: