I am running Isaac’s voice command detection sample application on Ubuntu 18.04. I trained a model for my own use case of voice command detection, and wanted to configure the sample application for this custom model. While configuring the json file of the sample app, I came across a config_file_path variable. What should I enter here? The training phase just output a .pb model file and a .json metadata file. Thanks!
Since you mentioned
config_file_path, I guess you are trying to use TensorflowInference Codelet. It is supposed to be the path of a Protobuf file as described here (https://docs.nvidia.com/isaac/isaac/packages/ml/ml.html?highlight=tensorflow#tensorflow-inference-chapter).
Not sure what framework you are using to train the model. If it’s tensorflow, TensorflowInference Codelet might help. For PyTorch, TorchInference Codelet might help. You may also convert the model to onnx and do inference with TensorRT. (https://docs.nvidia.com/isaac/isaac/doc/tutorials/nano.html?highlight=tensorrt)