search

lilanxiao / Rotated_IoU

Differentiable IoU of rotated bounding boxes using Pytorch
MIT License
416 stars 64 forks source link

About the 2D coordinates (x, y, w, h, alpha) #42

Closed KimSangYeon-DGU closed 2 years ago

KimSangYeon-DGU commented 2 years ago

Hi, I was curious that the x, y means the top left corner of a box. https://github.com/lilanxiao/Rotated_IoU/blob/3bdca6b20d981dffd773507e97f1b53641e98d0a/oriented_iou_loss.py#L9-L10

I'm asking because it seems some work treats the x, y as the center of a box.

lilanxiao commented 2 years ago

hi, thank you for the issue. Here x and y are the coordinates of the box center point. I chose this convention because most object detection methods (e.g. YOLO and R-CNN series) predict the center point instead of the upper left corner.

KimSangYeon-DGU commented 2 years ago

Aha, thanks for the clarification :)