TinyML based computer vision system

Abstract

Computer Vision is a field under artificial intelligence (AI) that translates digital images, videos and other visual inputs into useful and meaningful information. It often leverages machine learning and neural networks to process and analyze visual inputs efficiently. However, this process often requires high computation powers and is often run on high end processors for higher resolution and accuracy. In recent years, an emerging field called Tiny Machine Learning (TinyML) has surfaced and it focuses on deploying trained machine learning models on smaller, less powerful devices such as microcontrollers. This report discusses and explores the implementation of a TinyML-based computer vision system using the Arduino Nano 33 BLE Sense microcontroller and an OV5642 Arducam.