GestureToEmoji

Troubleshooting

• Why the loss does not go down? Arduino tinyml workshop gesture classification
• How to train/run Google Colab? Arduino Tinyml Workshop Gesture classification
• The loss from colab is always 0: you might only have one class, it needs to have at least 2 classes
• When collecting data from the serial monitior, uncheck "Show timestamp", so it doesn't include timestamp in your flex.csv file
• When compling "IMU_Classifier", "Error on compiling Arduino Nano 33 BLE", check the Arduino_TensorflowLite library version in your arduino IDE, it should be "2.1.0-ALPHA" without any "precomplied" label.
• Library version: Arduino_LSM9DS1 @1.1.0, Arduino_TensorflowLite @2.1.0-ALPHA

Demo Video

Forked from ArduinoTensorFlowLiteTutorials

Machine Learning on Arduino

TensorFlow Lite gesture training tutorial

In this tutorial we will teach a board to recognise gestures! We'll capture motion data from the Arduino Nano 33 BLE Sense board, import it into TensorFlow to train a model, and deploy a classifier onto the board using TensorFlow Lite for microcontrollers.

Credits

This tutorial is adapted from the workshop Sandeep Mistry, Arduino and Don Coleman, Chariot Solutions presented at AI/ML Devfest in September 2019.

Exercises

• Exercise 1: Development Environment
• Exercise 2: Connecting the Board
• Exercise 3: Visualizing the IMU Data
• Exercise 4: Gather the Training Data
• Exercise 5: Machine Learning
• Exercise 6: Classifying IMU Data
• Exercise 7: Gesture Controlled USB Emoji Keyboard
• Exercise 8: Next Steps