torchvision
torchvision copied to clipboard
[request] Non-maximum Suppression (NMS)
Hello! I would like to humbly make a request regarding NMS operator. I am currently working on porting this face detection and recognition implementation from py to R.
And to do so, NMS operator is required when boxing the detected faces.
I would be pleased to PR it myself, but it seems to be in C++ level and I felt insecure regarding autogens/best practices/compilers/stuff that you might reuse from torch devel toolkit itself.
I am open to be guided if needed to contribute with this. Also, I felt free to open a PR for boxes.py translation that is the task most beneficiated by NMS.
related links to what I'm talking about:
Thank you very much!