in the documentation for example models they have told to use *synpkg, binary_input.bin file and ndp_inference.py and put these files on to the board. For this project im not really sure how to create binary_input file and if we need ndp_inference.py or not. Also in the example models X.npy and Y.npy how are these generated?. if anyone could give some idea it would be great.
How do we go about accessing the hardware for interacting with the DL model? my project output feels like it is similar to this .
0:55-1:06
Syntiant Demos Mashup (in syntiant official yt channel, i cant post more links as i am a new user)
The information you are seeking related to Syntiant Moble Converter is covered with NDA. I believe your organization has NDA in place with us. Request you to reach out to your organization approver for a access to our customer support portal - https://syntiant.zendesk.com, and post the specific questions there and we will provide you more detailed response.
To give a high level view of your queries,
Syntiant Model converter provies ways to convert keras model (h5), & ONNX models to binary that is compatible to run on the Syntiant NDP hardware. You will have to save your model as one of those formats. A high level view of the hardware capability is listed in - https://www.syntiant.com/hardware . Depending on the hardware under consideration, the converted model will fit in.
The documentation of the model converter and the few examples shared in the packages provides informaiton on how to run the conversion utility and also the simulation tool, before running on the hardware.
The documentation of the model converter provides information.