[Feature Request] Extend transpose operator for MatND, Mat with channels and extend API #19751
Closed
4 tasks done
Labels
Milestone
Detailed description
In image preprocessing and postprocessing tasks for neural networks, the transpose operation is often used, for example, from the NumPy library:
img.transpose(1, 2, 0)
. Unfortunately, when we work in C++, we have no way to repeat this operation using OpenCV in one step. As a workaround, I use reshaping and memory copying, which reduces performance a lot.I think it is possible to implement this since this ability already exists in https://docs.opencv.org/master/d2/d3c/classcv_1_1dnn_1_1PermuteLayer.html
Steps to reproduce
Issue submission checklist
forum.opencv.org, Stack Overflow, etc and have not found solution
The text was updated successfully, but these errors were encountered: