Hi, I am trying to run a face recognition model from i.MX 8Mplus with npu. My environment setting is shown below.
- Hardware: NXP i.XM 8Mplus
- Software: Android version 12 (Tried 10 and 11 but failed)
Android SDK 33, NDK 21
- Model：insightface_quant (Downloaded WebFace600K_pfc .onnx weight file from https://github.com/deepinsight/insightface/tree/master/model_zoo and quantized into uint8 .tflite)
I used python 3.9 and tensorflow 2.11.0 for building tensorflow-lite via bazel. (Tried tf 2.4.0 first but failed as well). Both input and output type of quantized model is uint8.
I attached the error message, which is not really helpful for troubleshooting.
Both nnapi and xnnpack failed, while running on CPU without xnnpack succeeded.
Please help.
