Arduino TinyML project that uses a machine learning model to recognize digits in the camera feed, the model was trained using the MNIST dataset.
Source code can be found on GitHub or you can directly download Arduino project.
Link to the Edge Impulse project.
Demo
Tested Devices
The project has been tested on the following devices:
How to Run
Add the project to Arduino IDE using Sketch -> Include Library -> Add .ZIP Library...
.
Once the library has been added, go to File -> Examples
. You should see an entry within the list named martishin-mnist_inferencing
. Select it and click nano_ble33_sense -> nano_ble33_sense_camera
to load the project.
Use the Arduino IDE to build and upload the project. Once it is running, you can see inference results using Serial Monitor
:
Starting inferencing in 2 seconds...
Taking photo...
Predictions (DSP: 15 ms., Classification: 744 ms., Anomaly: 0 ms.):
Predictions:
0: 0.03906
1: 0.03125
2: 0.00391
3: 0.00000
4: 0.03125
5: 0.02344
6: 0.02734
7: 0.83594
8: 0.00391
9: 0.00391
In this case, the digit 7 was shown on the phone to the connected camera, the number after each digit is its score (confidence).