CornerNet: Detecting Objects as Paired Keypoints

CornerNet detects objects as a pair of keypoints— the top-left corner and bottom-right corner of the bounding box.

BibTex: