Embed Your Neural Networks with the Neural Compute...
summary
Shows how to run trained neural-network image-recognition models on edge hardware using the Intel Movidius Neural Compute Stick. Covers preparing TensorFlow/Keras models and deploying them to the NCS to meet constraints on latency and energy for mobile/embedded devices, illustrated with the AlphaShifumi image-recognition project from Zenika.
AI-written summary · about
chapters (11)
- 0:00Edge computing introduction
- 3:17Object detection and segmentation
- 5:45Neural network basics
- 12:37GPU vs Neural Compute Stick
- 14:26Neural Compute Stick overview
- 19:12Model development and prototyping
- 27:10Model compilation and profiling
- 31:33Model deployment and testing
- 35:33CPU vs Neural Compute Stick benchmark
- 37:03Transfer learning and challenges
- 40:18Conclusion and future trends
related talks
42mThe Stakes of Green AI for Embedded Systems
28mMachine Learning on the Edge
01h 02mNeural Networks: A Primer
13mAccelerating AI algorithms
37mDeep Learning Neural Network Acceleration at the Edge
26mBuild your Machine Learning Model on Edge with React Native
20mComputer Vision at the Edge with OpenVINO
07mMachine Learning on the Edge