Liao Peiyuan

Looking forward to the results! I implemented my own version of CIoU/DIoU loss but I couldn't obtain similar results on COCO when compared to huber loss.

I don't think my stats are comparable because I changed a lot of the original code.

Also, I haven't looked at your implementation, but AFAIK CIoU loss/DIoU loss only operates on xyxy/yxyx inputs, which is a lot different from EfficientDet's xy(log w)(log h)-relative format. How did...

Seems like what worked for me (on Anaconda) is to compile with 3.6.1, then reinstall 3.9 for running.

Sure!

I dug around and I think there isn't a easy work around. - I tried ```sys.module``` and xdist isn't listed - Then I tried some voodoo path parsing magic, and...

Removing the dim argument in your typedef should work. In this way, the dimension of the kd-tree will not be fixed at compile time.

Hi, thank you all for your interests in our work. It is true that we've omitted in the difference between the bpp we used with the one used in traditional...

this PR is ready for review & merge.

This PR is ready for review.