TinyML in Action: A Practical Guide to Machine Learning on Microcontrollers - Tapa blanda

Sinopsis

TinyML in Action is your hands-on guide to making that future real. This comprehensive book takes you from the foundations of embedded machine learning to full, deployable AI projects running on ultra-low-power devices. Whether you’re a developer, engineer, or curious maker, you’ll learn to design, train, and deploy efficient neural networks that live right on your hardware.

• Understand TinyML fundamentals: What it is, how it evolved, and why edge inference changes everything.

• Master the hardware–software ecosystem: Learn to choose the right microcontroller (ARM Cortex-M, ESP32, Arduino Nano 33 BLE Sense) and sensors for your application.

• Build and train real TinyML models: Use TensorFlow Lite for Microcontrollers, Edge Impulse, and CMSIS-NN to create compact, optimized neural networks.

• Deploy, debug, and optimize models on-device: Convert models to C-arrays, manage tensor arenas, and achieve real-time inference even on devices with <256 KB RAM.

• Implement power-efficient designs: Learn duty cycling, quantization-aware training, and firmware optimization for long battery life.

• Develop real-world edge AI projects: Gesture recognition, keyword spotting, image detection, predictive maintenance, and environmental monitoring—all step-by-step.

You’ll walk through the entire TinyML workflow, from data → model → deployment, using practical, real-world examples grounded in official TensorFlow Lite Micro and Arduino references. Each chapter builds on the previous with structured learning: theory, implementation, optimization, and testing. You’ll also find dedicated troubleshooting sections, hardware setup guides, and power-profiling strategies for dependable edge-AI performance.

By the end of this book, you’ll know how to:

• Collect and preprocess sensor data directly on your board.

• Train compact neural networks using Python and TensorFlow/Keras.

• Quantize, prune, and compress models for memory-limited devices.

• Flash the compiled model and run inference in real time.

• Profile latency, RAM usage, and power consumption with confidence.

• Scale your TinyML applications with OTA updates and cloud integration via MQTT, AWS IoT, or Azure IoT Hub.

This book is perfect for:

• Embedded developers exploring AI for the first time.

• Machine learning practitioners looking to deploy models at the edge.

• IoT engineers building intelligent sensors, wearables, or industrial monitors.

• Students, educators, and makers passionate about sustainable, low-power AI.

No prior deep learning expertise is required — every example is practical, commented, and reproducible.

• Gesture recognition using an IMU sensor.

• Keyword spotting wake-word detector.

• Person detection on an ESP32-CAM.

• Predictive maintenance system with vibration data.

• Smart environmental monitor fusing sound, temperature, and motion.

Each project reinforces your understanding of embedded AI optimization, ensuring you can design models that think, sense, and respond — all within the constraints of a microcontroller.

Start reading TinyML in Action today.

"Sinopsis" puede pertenecer a otra edición de este libro.