154
click to vote
CVPR
2011
IEEE
14 years 11 months ago
2011
IEEE
The bounding box representation employed by many popular object detection models [3, 6] implicitly assumes all pixels inside the box belong to the object. This assumption makes th...