I did not find any reference for Caffe and TF parsers in PyArmnn. I would like to know how to load Caffe/TF models and run inference. Can anyone help me?
Hello,
You can find an example with Tflite here: https://developer.arm.com/solutions/machine-learning-on-arm/developer-material/how-to-guides/accelerate-ml-inference-on-raspberry-pi-with-pyarmnn/run-ml-inference-with-pyarmnn
and also here: https://github.com/ARM-software/ML-examples/tree/master/pyarmnn-fire-detection
Best,
Mina
Hello, these examples both show how to load a tflite model, but the Caffe and TF parsers are different as they accept a different number of parameters. Therefore I cannot rely on these examples.
Yes, you are right. There are no examples with Caffe and Tensorflow and PyArmNN. There are some with ArmNN and C++ in the same repository.
Unfortunately, there are no plans to add more examples with Caffe or Tensorflow at the moment.
Is converting your model to Tflite an option?
Also, note that in 20.11 release Tflite delegate was added as well as support for more operators: https://github.com/ARM-software/armnn/releases