I had a problem when I tried to inference with my onnx IR documents. Before this, I've already transferred my onnx network. The network was firstly built on MATLAB Deep Network Designer, then I used a transfer formulation to transfer it to onnx format. When I tried to transfer it to IR documents in OpenVINO, it always came out an error that something was wrong with the input_shape and I posted a question on this community. However, now the input data is prepared but a new error came out. The input data was a file saved by GNU Radio file sink. At first, it was a binary file but I've already transferred it to a two dimension data file because the input data in MATLAB was this format.
My operating environment:
Thanks for reaching out. I have tested your model and encountered the same error. I tested your model with the OpenVINO Benchmark app and the same error arises. It seems that your onnx model might miss some parameters during the conversion. Could you share your command for converting the model into the IR? Did you specify the batch parameter?
Thank you for your question. If you need any additional information from Intel, please submit a new question as this thread is no longer being monitored.