Faster RCNN

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks^[1]
Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun
arXiv:1506.01497

official github
tensorflow ver.
py faster rcnn github

(objectness score를 학습하는) RPN(Region Proposal Network)을 제안한다.

Fast RCNN에서 보았듯, region proposal computaion이 (test time의) 병목이다. CNN으로 이루어진 RPN을 학습시켜서 이미지 한장당 300개정도의 proposal만으로 state-of-the-art accuracy를 얻는다. RPN은 기존의 Fast RCNN과 parameter를 share한다.

==x==

=x=

↑ Ren, Shaoqing, et al. "Faster R-CNN: Towards real-time object detection with region proposal networks." Advances in neural information processing systems. 2015.

[1] Ren, Shaoqing, et al. "Faster R-CNN: Towards real-time object detection with region proposal networks." Advances in neural information processing systems. 2015.

[1]