This is a PySide6 desktop application that connects to an M5StickC Plus device over Bluetooth Low Energy (BLE). It allows you to stream, record, and segment IMU sensor data (accelerometer + gyroscope) for activity recognition, TinyML, or time-series analysis tasks.

You can use this tool for real-time sensor data logging, data labeling, and TinyML dataset preparation.

You can find fully built versions for Windows, macOS, and Linux in the GitHub Actions build artifacts:

👉 Download from GitHub Actions (latest build)

📦 Artifacts include .exe, .dmg, and .AppImage files ready to run on your platform.

Once downloaded and installed, you’ll see a screen like this:

Before collecting data, flash your M5StickC Plus with this firmware:

👉 View Arduino sketch (Gist)

This firmware:

• Streams raw IMU packets over BLE
• Computes and sends feature vectors every second
• Includes BLE UUIDs that the app listens to

1. Power on your M5StickC Plus after flashing the firmware.
2. Open the desktop app.
3. Click 🔍 Scan.
   • You’ll see your M5Stick device (e.g., M5Stick-IMU) in the list.
4. Select the device and click ✅ Connect.
5. Once connected, click ⏺ Start Recording.
   • This will begin recording both raw data and feature vectors into two separate CSV files.
6. During recording, you can use Segment 1 / Segment 2 / Segment 3 buttons to label different activity phases (e.g., Walking, Sitting, Running).
   • These segment labels will be recorded alongside your data in the CSV.
7. When done, click ⏹ Stop Recording.
8. Click 🔌 Disconnect when you're finished.
   • You can reconnect again anytime and repeat the process.

• raw_data_*.csv → Contains raw IMU data: 10x6 int16 values per packet
• features_*.csv → Contains feature rows with segment labels
• Both include timestamps and segment markers for easy training

This mode allows you to view live predictions from a TinyML model running directly on the M5StickC Plus.

1. Flash your M5StickC Plus with the TinyML prediction firmware.

   👉 View TinyML BLE Arduino Code (Gist)

   This firmware runs a trained Tiny ML model on raw IMU data, and sends predictions like "Still" or "Moving" over BLE.

2. Launch the App (same one used for data collection).

3. Click 🔍 Scan to find your device (TinyML-M5), and then click ✅ Connect.

4. Once connected, scroll down and click the 👁 Show Output button.

   This will activate live BLE updates from the device’s prediction characteristic.

Below is an example of what the GUI displays when predictions are being streamed from the M5StickC Plus:

The output updates every 0.5 seconds and shows predicted labels such as “Still”, “Moving”, etc.

This app was developed and built on macOS, specifically for M-series (Apple Silicon) devices.

If you try to run it the usual way (e.g., from GitHub Actions builds for Windows/Linux/MacOS), you may run into macOS security restrictions — gatekeeper, app signing, notarization, etc.

✅ For convenience, I’ve already built the app locally and placed the full macOS build inside the dist/ folder.

Inside dist/M5StickDataLogger/, you’ll find:

• M5StickDataLogger (app folder)
• M5StickDataLogger (Unix executable file) ← Run this to launch the app

🧩 You might need to:

• Enable Bluetooth permissions for the app in System Settings
• Restart the app once after first launch

If you're familiar with macOS app builds and permissions, feel free to try running directly from GitHub Actions too. But for most users, the local dist/ build should work out of the box.

Thank you and enjoy logging your M5Stick data!