Hierarchical objectness network for region proposal generation and object detection

作者：

Highlights：

• We propose a hierarchical objectness network for region proposal generation and object detection to address the inaccurate localization problem.

• We subtly localize the objects by predicting the stripe objectness, i.e., a group of probabilities reflecting the existence of the object in each location of the candidate proposal.

• We construct the hierarchical features by reversely connecting multiple convolutional layers to detect objects with large-scale variations.

摘要

•We propose a hierarchical objectness network for region proposal generation and object detection to address the inaccurate localization problem.•We subtly localize the objects by predicting the stripe objectness, i.e., a group of probabilities reflecting the existence of the object in each location of the candidate proposal.•We construct the hierarchical features by reversely connecting multiple convolutional layers to detect objects with large-scale variations.

论文关键词：Object detection,Object localization,Region proposal generation,Convolutional neural network

论文评审过程：Received 10 January 2018, Revised 25 April 2018, Accepted 13 May 2018, Available online 5 June 2018, Version of Record 18 June 2018.

论文官网地址：https://doi.org/10.1016/j.patcog.2018.05.009