Does anyone know whether YOLOv4 and mAP metric would be implemented in Fastai v1 or v2?
mAP has been. Check out the thread discussing v2 object detection.
Is it this one? Mean average precision (mAP)
Thanks. How about YOLOv4? any plans for that?
Check out the discussion here:
If there’s a
PyTorch implementation, you can use it. The challenge is getting the rest of it working too, which is also a problem with integrating it itself into the library. There’s too large of a difference between the models IMO and the support needed for them to have them all on hand like
ResNet. However, that doesn’t mean it can’t be done