Quick step to identify the input nodes of a neural network model.
Converted ONNX model to OpenVINO™ IR format using Model Optimizer command:
--input_model "{onnx_path}
--input_shape [1,512, 896, 3], [1,512, 896, 4]
--input frame, prob
--data_type FP16
--output_dir "{model_path.parent}"
Received error:
[ ERROR ] Exception occurred during running replacer "REPLACEMENT_ID" (<class 'extensions.front.user_data_repack.UserDataRepack'>): No node with name frame
The Model Optimizer did not recognize the input nodes frame and prob, due to wrongly specifying the input nodes of the model.
- Check the correct input nodes for the model using Netron.
- Change the input nodes option from --input frame, prob to --input 0,1.