Machine Learning on the edge using TensorFlow Lite
summary
Shows converting a pre-trained image classifier to TensorFlow Lite then deploying it on Raspberry Pi and microcontrollers, using the TensorFlow Lite converter and running the interpreter. Demonstrations use MobileNet (ImageNet) for image classification—examples return Labrador 27%, tiger 96%—and MobileNet SSD (COCO) for object detection—examples report bottle ~55%, remote control ~53% and person detection. Mentions a Nuru cassava app trained on 5000 images and tiny models that fit 384K RAM microcontrollers.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 1:10What is machine learning
- 5:02TensorFlow basics
- 6:27TensorFlow Lite overview
- 9:18Image classification demo
- 15:44Object detection demo
- 22:01Real-world applications
- 24:13Microcontroller use cases
- 26:42Q&A and closing
related talks
39mLeveraging the power of C++ for efficient machine learning on embedded devices
48mLeveraging the Power of C++ for Efficient Machine Learning on Embedded Devices
34mTensorFlow Lite
01h 01mML and the IoT: Living on the Edge
32mAI for Mobile and IoT Devices: TensorFlow Lite
27mMachine Learning and Mobile
55mBuilding with TensorFlow Lite for microcontrollers
39mMachine Learning on Your Device: The Options