search

deepinsight / insightface

State-of-the-art 2D and 3D Face Analysis Project
https://insightface.ai
22.96k stars 5.36k forks source link

onnx_helper error #2300

Open SuyueLiu opened 1 year ago

SuyueLiu commented 1 year ago

when I run onnx_helper to check the trained onnx model(generate by the default code), I got an error: image

Traceback (most recent call last): File "onnx_helper.py", line 251, in err = handler.check(args.track) File "onnx_helper.py", line 166, in check batch_result = self.check_batch(test_img) File "onnx_helper.py", line 200, in check_batch net_out = self.session.run(self.output_names, {self.input_name: blob})[0]

onnxruntime.capi.onnxruntime_pybind11_state.InvalidArgument: [ONNXRuntimeError] : 2 : INVALID_ARGUMENT : Non-zero status code returned while running Gemm node. Name:'Gemm_185' Status Message: GEMM: Dimension mismatch, W: {512,512} K: 16384 N:512

input: [None, 3, 112, 112] The onnx_helper blob shape: [32, 3, 112, 112]

Does any one know how to solve this problem?

nttstar commented 1 year ago

Have you converted the ONNX model to a format that supports batch inference? (where input_shape[0] is in string type)

SuyueLiu commented 1 year ago

Have you converted the ONNX model to a format that supports batch inference? (where input_shape[0] is in string type)

yes I did image

hyperf0cus commented 7 months ago

Any updates???

hyperf0cus commented 7 months ago

Did you solve this error? Why ignoring?