Example of live gesture inference with Particle Photon 2

This Particle Workbench project demonstrates how to perform live inference with an accelerometer using the Particle Photon 2.

Hardware connections

You will need the following:

• ADXL362 accelerometer breakout board
• Particle Photon 2
• Solderless breadboard
• Jumper wires
• USB micro-B cable

Note that you can also purchase the Particle Edge ML kit.

Follow the directions on this Edge Impulse tutorial to connect the ADXL362 accelerometer to the P2. The P2 pinout diagram can be found here.

ADXL362	Photon 2
VIN	3V3
GND	GND
SCL (SCK)	SCK (D17)
SDA (MOSI)	MO (D15)
SDO (MISO)	MI (D16)
CS	A2 (D13)
INT1	Not connected
INT2	Not connected

Important! You must register your P2 on your Particle account before continuing

Plug your P2 into your computer. Head to https://docs.particle.io/device-setup/ and follow the instructions to register it.

Repository contents

The source code for this example can be found in the src/ directory. You probably want to look at src/main.cpp. This contains the application demo for performing live inference by reading from the ADXL362 accelerometer.

Note that this repository has a pre-trained model that has been deployed from Edge Impulse. The model was trained by using a small dataset of continuous gestures. You can clone and deploy the model from this public Edge Impulse project.

There is also a static buffer example found in examples/static_buffer/main.cpp. Feel free to look at that demo to see how you might test inference without connecting the accelerometer.

Build and flash

Install the Particle Workbench by following these instructions. We recommend simply installing the VS Code extension.

Once installed, click the Particle Workshop extension on the left-side toolbar of VS Code. Alternatively, you can press ctrl + shift + p (Win/Linux) or cmd + shfit + p (Mac) to bring up the command palette and enter >Particle: to access the various Workbench commands.

1. In Workbench, select Particle: Import Project and select the project.properties file in the directory that you just downloaded and extracted.

2. Use Particle: Configure Project for Device and follow the prompts:

   a. Select [email protected]

   b. Choose Photon/P2 as your target

   c. Go to https://console.particle.io/devices and copy the ID for your P2 device. Paste it into the prompt when asked Device ID or name?

3. Compile with Particle: Compile application (local)

4. Flash with Particle: Flash application (local)

5. Open the serial monitor with Particle: Serial Monitor to view the inference output

Important! At this time you cannot use the Particle: Cloud Compile or Particle: Cloud Flash options; local compilation is required.

Deploy your own model

If you wish to deploy your own model, deploy with the Particle target from the Deployment page in Edge Impulse. Alternatively, you can deploy with the C++ library option and copy the following folders to the src/ directory in this project:

• edge-impulse-sdk/
• model-parmaeters/
• tflite-model/

When deployed, these folders contain your trained model and all the necessary code to run inference.