Confusion About Pixel Coordinates

[kkaytekin]

I am a little confused about whether pixel coordinates represent the pixel centers or the pixel corners.

• If the pixel coordinates represent pixel corners, rescaling image is simply multiplying the first two rows of the intrinsic matrix, as done in human_loader.py
• But looking at the implementation of depth2pts here shows that pixel coordinates represent the pixel centers. In this case, the pixel coordinates need to be transformed another way, as shown in this Stackoverflow Post

I'd appreciate it if you could explain which is the case here.

Best,