Overview
Save Big on Coursera Plus. 7,000+ courses at $160 off. Limited Time Only!
Explore the cutting-edge developments in ultra-low power always-on computer vision at Qualcomm in this tinyML Talks Morocco presentation. Delve into the challenges of achieving computer vision at 1mW for TinyML applications and discover the innovative solutions across sensor technology, custom ASIC components, architecture, algorithms, software, and trainable models. Learn about Qualcomm Technologies' always-on computer vision module, featuring a low-power monochrome qVGA CMOS image sensor and an ultra-low power custom SoC. Understand how challenging traditional computer vision assumptions is enabling new applications in mobile phones, wearables, and IoT. Gain insights into the Qualcomm QCC112 chip, its potential use cases, and the overview of training tools for always-on computer vision systems. Follow the presentation's structure, covering topics such as tinyML background, use cases, adaptive compute, human and person detection, gesture recognition, lighting considerations, and ML tools like Tuner and AI Engine.
Syllabus
Introduction
Introducing the speaker
What is tinyML
Background
Use Cases
Traditional Approaches
Adaptive Compute
Human Detection
Person Detection
Gesture Detection
Lighting
Phase Detection
Image Examples
Latency Requirements
Demo
ML Tools
Tuner
AI Engine
QA
Conclusion
Taught by
tinyML