Teachable Machine Boilerplate

Overview

Teachable Machine Boilerplate is a small boilerplate project that showcases the use of tensorflow.js to create projects similar to Teachable Machine. The project demonstrates how to create a KNN classifier that can be trained in real-time using the webcam image. The code is designed to be simple and serves as a starting point for new projects. Behind the scenes, the project utilizes the MobileNet activation to process the webcam image. By using the high-level abstractions learned by the MobileNet network, the boilerplate allows users to train the KNN classifier with minimal samples, enabling recognition of specific classes such as smiles vs frowns or small body movements. This technique is known as Transfer Learning.

Features

• Create a KNN classifier using tensorflow.js
• Train the classifier in real-time using the webcam image
• Utilize the MobileNet activation for processing the webcam image
• Leverage Transfer Learning to recognize specific classes with minimal samples

Summary

Teachable Machine Boilerplate is a simple project that demonstrates the use of tensorflow.js to create projects similar to Teachable Machine. By leveraging the MobileNet activation and Transfer Learning, users can train a KNN classifier in real-time using the webcam image. The code provides a starting point for new projects and enables the recognition of specific classes with minimal training samples.