How to use Pytorch Tensor object in Opencv without convert to numpy array?

Question

I’m trying to develop a text detection application with Pytorch and Opencv in Python. I can use Pytorch tensor with Opencv like below.

val = y[0,:,:,0].data.cpu().numpy()
cv2.threshold(val , 0.4, 1, 0)

But it takes a lot of time. I need to do this operation by using the tensor object. How can I do that?

Berriel · Accepted Answer · 2020-01-07 12:25:12Z

4

Given that the last 0 in your threshold call means cv.THRESH_BINARY, it follows this function:

As your maxval is set to 1, you can replace this threshold call with something like this:

(y[0,:,:,0] > 0.4).float()

I am casting to float, but you can change that as you wish, ofc. Or even to something like:

(y[0,:,:,0] > 0.4).to(dtype=y.dtype)

so that it will remain with the same data type.

answered Jan 7, 2020 at 12:25

Berriel

13.8k4 gold badges51 silver badges73 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

Emre BEGEN Over a year ago

Thank you four your answer. I think it can work but there are more opencv functions in my code and with your way I have to change all of these methods. I think It would be better if I know how I can give tensor data to opencv methods.

Berriel Over a year ago

@EmreBEGEN Without converting to numpy, you won't be able to do it, and you shouldn't need it. Walk us through your use case and we may be able to give you an advice.

Collectives™ on Stack Overflow

How to use Pytorch Tensor object in Opencv without convert to numpy array?

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related