Book Synopsis & Overview
Practical TinyML and ONNX Runtime deployment on resource-constrained embedded devices for offline computer vision and audio classification.
Chapter 1: 8-Bit and 4-Bit Integer Quantization
Converting floating-point weights to INT8 and INT4 representations allows complex vision models to execute within 2MB SRAM footprints at 30 FPS.
Publishing & Format Specifications
- • Format: Reflowable EPUB 3.0 & Print-Ready PDF
- • DRM: 100% DRM-Free Lifetime Ownership
- • Compatibility: Kindle, Apple Books, Kobo, Android, and Web Reader