Parsing custom ONNX - YoloV7

Community Projects
1

Hello Hailo,

I have an error when I try to parse my custom yolov7 model.

I know it exists in Hailo Zoo but I’d like to use my custom models because I plan to use other custom models not available in Hailo Zoo.

I run under Almalinux 8 (RHEL) under an Ubuntu 22 docker. I successfully installed the hailo dataflow compiler for my hailo8 card (hailo_dataflow_compiler-3.33.0-py3-none-linux_x86_64.whl) but when I try to parse, I receive this error :

root@693ec62ef232:/# hailo parser onnx /models/model_inferred.onnx
[info] No GPU chosen and no suitable GPU found, falling back to CPU.
[info] Current Time: 15:23:20, 10/17/25
[info] CPU: Architecture: x86_64, Model: Intel(R) Core™ i7-10875H CPU @ 2.30GHz, Number Of Cores: 16, Utilization: 5.1%
[info] Memory: Total: 15GB, Available: 8GB
[info] System info: OS: Linux, Kernel: 4.18.0-553.77.1.el8_10.x86_64
[info] Hailo DFC Version: 3.33.0
[info] HailoRT Version: Not Installed
[info] PCIe: No Hailo PCIe device was found
[info] Running hailo parser onnx /models/model_inferred.onnx
[info] Translation started on ONNX model model_inferred
[info] Restored ONNX model model_inferred (completion time: 00:00:00.37)
[info] Extracted ONNXRuntime meta-data for Hailo model (completion time: 00:00:01.23)
[info] Simplified ONNX model for a parsing retry attempt (completion time: 00:00:03.68)
Traceback (most recent call last):
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/sdk_backend/parser/parser.py”, line 235, in translate_onnx_model
parsing_results = self._parse_onnx_model_to_hn(
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/sdk_backend/parser/parser.py”, line 316, in _parse_onnx_model_to_hn
return self.parse_model_to_hn(
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/sdk_backend/parser/parser.py”, line 359, in parse_model_to_hn
fuser = HailoNNFuser(converter.convert_model(), net_name, converter.end_node_names)
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/model_translator/translator.py”, line 82, in convert_model
self._create_layers()
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/model_translator/edge_nn_translator.py”, line 39, in _create_layers
self._add_direct_layers()
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/model_translator/edge_nn_translator.py”, line 121, in _add_direct_layers
self._layer_callback_from_vertex(vertex)
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/model_translator/onnx_translator/onnx_translator.py”, line 434, in _layer_callback_from_vertex
self._create_ew_add_layer(vertex)
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/model_translator/onnx_translator/onnx_translator.py”, line 1156, in _create_ew_add_layer
self._create_ew_layer(vertex, EWAddLayer)
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/model_translator/onnx_translator/onnx_translator.py”, line 1179, in _create_ew_layer
const_layer, ew_op_input = self._create_ew_const_input_layer(vertex)
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/model_translator/onnx_translator/onnx_translator.py”, line 1316, in _create_ew_const_input_layer
const_layer = ConstInputLayer.create(f"{vertex.name}_input", [const_output_shape], input_values)
UnboundLocalError: local variable ‘const_output_shape’ referenced before assignment

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File “/hailo_dataflow_compiler_venv/bin/hailo”, line 8, in 
sys.exit(main())
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/tools/cmd_utils/main.py”, line 111, in main
ret_val = client_command_runner.run()
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/tools/cmd_utils/base_utils.py”, line 68, in run
return self._run(argv)
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/tools/cmd_utils/base_utils.py”, line 89, in _run
return args.func(args)
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/tools/parser_cli.py”, line 212, in run
self._parse(net_name, args, tensor_shapes)
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/tools/parser_cli.py”, line 298, in _parse
self.runner.translate_onnx_model(
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_common/states/states.py”, line 16, in wrapped_func
return func(self, *args, **kwargs)
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/runner/client_runner.py”, line 1192, in translate_onnx_model
parser.translate_onnx_model(
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/sdk_backend/parser/parser.py”, line 276, in translate_onnx_model
parsing_results = self._parse_onnx_model_to_hn(
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/sdk_backend/parser/parser.py”, line 316, in _parse_onnx_model_to_hn
return self.parse_model_to_hn(
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/sdk_backend/parser/parser.py”, line 359, in parse_model_to_hn
fuser = HailoNNFuser(converter.convert_model(), net_name, converter.end_node_names)
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/model_translator/translator.py”, line 82, in convert_model
self._create_layers()
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/model_translator/edge_nn_translator.py”, line 39, in _create_layers
self._add_direct_layers()
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/model_translator/edge_nn_translator.py”, line 121, in _add_direct_layers
self._layer_callback_from_vertex(vertex)
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/model_translator/onnx_translator/onnx_translator.py”, line 434, in _layer_callback_from_vertex
self._create_ew_add_layer(vertex)
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/model_translator/onnx_translator/onnx_translator.py”, line 1156, in _create_ew_add_layer
self._create_ew_layer(vertex, EWAddLayer)
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/model_translator/onnx_translator/onnx_translator.py”, line 1179, in _create_ew_layer
const_layer, ew_op_input = self._create_ew_const_input_layer(vertex)
File “/hailo_dataflow_compiler_venv/lib/python3.10/site-packages/hailo_sdk_client/model_translator/onnx_translator/onnx_translator.py”, line 1316, in _create_ew_const_input_layer
const_layer = ConstInputLayer.create(f"{vertex.name}_input", [const_output_shape], input_values)
UnboundLocalError: local variable ‘const_output_shape’ referenced before assignment

Following some of the directives found here Onnx to Har parsing error
I tried to remove any unknown shape using the following command, but it didn’t help. From what I could see, the models has no dynamic shape.

python -m onnxruntime.tools.symbolic_shape_infer --input /models/original_model.onnx --output /models/model_inferred.onnx --auto_merge

Am I missing something in the parsing process ?
Thank you !

2

Hi @K_E, The Dataflow Compiler is designed to be used only on X86 systems at the moment, it might be the issue, I;m not sure since it’s not tested.

Have you tried using the model-zoo recipe for converting the yolov7?

3

Thank you for your reply !

I am not sure to understand your remark : I am running on a x86 system, just a different operating system (Almalinux with is RHEL based VS Ubuntu).

I have tested the yolov7 from Hailo-Zoo to make sure the card is installed properly and it seems to be ok, but in the end, I need to transform my custom ONNX because we have some intern models that won’t be available in the Zoo.

4

you’re right, my bad, not sure what led me to belive this an ARM based machine.
In any case, if you use a model that is supported ni the model zoo, but your mocel is pruned, or have less classes, or different dimensions etc. The model-zoo can still be your friend! We have API to change the conversion to comply with these changes.

5

After adding some prints in the hailo code, it turns out that the onnx should have an input batch size of 1 to work properly.

I changed and it now works.

1 Like