How I can use my model of AI in tflite format on arduino or esp, and what i need?

TensorFlow General Discussion
,
1

What files or model format i need to use model of object detection on esp or arduino, it would be good, if someone can give example of code

3

I haven’t gone to Arduino level yet, because Coral.AI is has an A53 ARM CPU, a GPU and a TPU as well, I’ve run TFlite on that.
However looking in the docs it seems LiteRT LiteRT for Microcontrollers  |  Google AI Edge  |  Google AI for Developers could be a good direction. One of your limiting factors will be memory, LiteRT seems to support qunatization which will help. But for object detection (what I did with Coral) you may need to resort to smaller models, or certainly train them for your specific task.

Arduino has some docs about TFLite: https://docs.arduino.cc/tutorials/nano-33-ble-sense/get-started-with-machine-learning/