Call Session class method name Run failed,don't know why

[angel7039]

I got onnx model which is dynamic shape, when use this model to infererence, there is an error during call session.run method,I don't why,somebody can tell me?

the C++ code is like this,

###to get input data:
this->input_image_.resize(resizeH * resizeW * resizedImage.channels());
for (int c = 0; c < 3; c++)
{
for (int i = 0; i < resizeH; i++)
{
for (int j = 0; j < resizeW; j++)
{
float pix = img.ptr(i)[j * 3 + 2 - c];
this->input_image_[c * resizeH * resizeW + i * resizeW + j] = pix / 255.0;

		}
	}
}

###create tensor data;
array<int64_t, 4> input_shape_{1, 3, input_h, input_w };
auto allocator_info = MemoryInfo::CreateCpu(OrtDeviceAllocator, OrtMemTypeCPU);
Value input_tensor_ = Value::CreateTensor(allocator_info, input_image_.data(), input_image_.size(), input_shape_.data(), input_shape_.size());

###the error occurs this method;
vector ort_outputs = ort_session->Run(RunOptions{ nullptr }, &inputs, &input_tensor_, 1, &outputs, 1);

Document Details

• Title: Reference
• Page: https://onnxruntime.ai/docs/reference/
• Page Source: index.md

[pranavsharma]

Please paste the error you see + model + repro code.

[angel7039]

Please paste the error you see + model + repro code.

onnx_test.zip

please check from the attach file.

the error seems like this :

[yuslepukhin]

You are experiencing an invalid memory access, not an Onnxruntime error. You can see in the debugger what location you are trying to read, how you got there. Most likely your shape and your buffers are not in agreement.