For my first trials I decided to use the TinyML kit: which contains Arduino Nano 33 BLE Sense and the OV7675 camera (kit).

I use the following libraries on arduino IDE:

• Arduino Mbed OS Nano Boards v3.0.1
• Arduino_OV767X v0.0.2
• Arduino_TensorFlowLite (from repository) v latest in the repository

There is another library available as well: Harvard_TinyMLx (pepared for TinyML course available on EDX) - which contains the library for the camera and tensorflow. However, I wanted to use to official ones as this would enable me easier comparison of the results with Edge Impulse and help me avoid unexpected issues when searching through available examples.

In the Arduino_OV767X library there is an example available to retrieve the image - which is very useful to make sure the hardware works.
The Arduino_TensorFlowLite library also comes with several interesting examples, even person recognition - however I am interested to do a full chain - hopefully with the same (or perhaps even better..?) results.