In i.MX Machine Learning User's Guide LF5.15.5-1.0.0, I found the following release note sentence:
NNAPI Delegate of TensorFlow Lite and NNAPI Execution Provider of ONNX Runtime is deprecated and will be removed in the future. For leveraging ML model acceleration use VX Delegate instead.
The nnapi is maintained by Google and VX Delegate is maintained by NXP. The maintenance of nnapi is inconvenient.
Thank you for your reply.
I understand that NNAPI is very difficult to maintain.