The aim of this comparison is to compare these two technologies used in Machine Learning applications: Genuino 101 vs uTensor. In particular we are going to compare the Intel Curie Module (Pattern Matching Engine) and TensorFlow.
2. ● Microcontroller Intel Curie
● Flash Memory 196 kB
● SRAM 24 kB
● Clock Speed 32MHz
● I/0 communication Bluetooth LE
● Sensors 6-axis
accelerometer/gyro
uTensor is a machine learning inference framework built on Mbed
and Tensorflow. It converts ML models to C++ source files, ready
to be imported into MCU projects.
4. TensorFlow is an open source software library for machine
learning, which provides tested and optimized modules
useful in the implementation of algorithms for different types
of perceptive tasks and language comprehension.
Operating systems supported by TensorFlow:
● Windows
● Linux
● MacOS
● Android
TensorFlow provides native APIs in:
● Python
● C/ C ++
● Java
● Go
● RUST
Intel Curie module is a hardware product offering design
flexibility in a small form factor.
● Low-power solution
● Motion sensor
● Bluetooth low energy
Pattern matching capabilities for optimized analysis of sensor
data-enabling quick and easy identification of actions and
motion
5. Comparison based on three main points:
● Available models
● Hardware requirements and performances
● Easy of use
Comparison overview
vs
7. Hardware requirements and
performances
● Related to Genuino 101 hardware limits:
○ Flash Memory 196 kB
○ SRAM
24 kB
● Does not depend on hardware limits
Results
High AccuracyLow Accuracy
8. Easy of use
Training Prediction Training Prediction
Local training External Training
Training can be done directly on board
Minor fields of use
Less intuitive
Greater potential
Load Model
on board
9. Summary
● To build simple Machine Learning
based projects
● Low cost hardware solution
● Easy implementation
● Low energy consumption
● To build complex Machine Learning based
projects
● Select hardware according to needs
● Better accuracy results