C++ tensor operation

C++
1

I have a tensor which contains boxes predict by detection. and it’s format like:

cx, cy, w, h

I want convert this tensor to:

x1, y1, w, h

So I can do this in python:

boxes[:, :2] -= boxes[:, 2:] / 2
    boxes[:, 2:] += boxes[:, :2]

But how to do this in libtorch???

support I have

torch::Tensor boxes
2

Are you looking for boxes.slice or boxes.narrow?

Best regards

Thomas

3

thank u for your reply Tomas.

Can you elibrate your answer? I am not quit understand of the narrow usage. I know there is a slice method of tensor.

I got trouble in slice.

My tensor is size as : [2976, 4], which is 2976 rows and 4 columns.
I want get the first 2 columns:

priors.slice(1, 0, 2)

Got error:

terminate called after throwing an instance of 'std::runtime_error'
  what():  invalid argument 0: Sizes of tensors must match except in dimension 0. Got 1 and 2 in dimension 1 at /media/jintain/sg/ai/libs/pytorch/aten/src/THC/generic/THCTensorMath.cu:71

Did I slice it wrong?