CornerNet: Detecting Objects as Paired Keypoints

Hei Law, Jia Deng; Proceedings of the European Conference on Computer Vision (ECCV), 2018, pp. 734-750


We propose CornerNet, a new approach to object detection where we detect an object bounding box as a pair of keypoints, the top-left corner and the bottom-right corner, using a single convolution neural network. By detecting objects as paired keypoints, we eliminate the need for designing a set of anchor boxes commonly used in prior single-stage detectors. In addition to our novel formulation, we introduce corner pooling, a new type of pooling layer that helps the network better localize the corners. Experiments show that CornerNet achieves a 42.1% AP on MS COCO, outperforming all existing one-stage detectors.

Related Material

[pdf] [arXiv]
author = {Law, Hei and Deng, Jia},
title = {CornerNet: Detecting Objects as Paired Keypoints},
booktitle = {Proceedings of the European Conference on Computer Vision (ECCV)},
month = {September},
year = {2018}