Simple Neural Network Library for TinyCLR (MNIST Demo)

TinyCLR OS
1

20210811_090021
20210811_0900213024×4032 1.83 MB

This is a ported of light-weight NN library for TinyCLR. Take a look the demo.
(47) TinyCLR - Neural Network Demo (Mnist) - YouTube

you can get the source with sample implementation from
mifmasterz/TinyCLR-NeuralNetwork: This is experiment how to run simple neural network to recognize hand written digit (MNIST) (github.com)
With this library you will be able to:

  1. training your model on pc
  2. save / load your model
  3. do model inference on device

Note: Make sure you enable the external RAM on the device, the built-in RAM won’t be enough to load the model: External Memory (ghielectronics.com)

Hope you like it…
"I Hope someday GHI will support TFLite format to do DL task on TinyCLR device. "

4 Likes
2

This is great to see and can’t wait to play with it. I wonder if there is a commercial use for it

3

-predictive maintenance
-anomaly detection
-smart robotic
-object detection
-etc

GHI should considered to add this library to their official library: Understand the C++ library  |  TensorFlow Lite