The YOLOv8 model is trained to search for aircraft on the ground. This was done at Colab

After that, it was exported to tflite format and launched on google coral. It works.

https://coral.ai/docs/accelerator/get-started

for all files path

python3 ./detect_airplanes_coral.py --model ./best_full_integer_quant_edgetpu.tflite --input ~/googledrive/neuro/airbus/extras

or for one file

python3 ./detect_airplanes_coral.py --model ./best_full_integer_quant_edgetpu.tflite --input ./495b73c8-024f-46cc-b426-05e49bbe5074.jpg

or without coral TPU

python3 ./detect_airplanes_tflite.py --model ./best_float32.tflite --input ./495b73c8-024f-46cc-b426-05e49bbe5074.jpg

Coral

TfLite