DOOBNet: Deep Object Occlusion Boundary Detection from an Image

TL;DR

This paper introduces DOOBNet, a deep multi-task network with a novel loss function for improved object occlusion boundary detection, addressing class imbalance and achieving state-of-the-art results.

Contribution

The paper presents a unified end-to-end network with an attention loss to handle class imbalance and multi-scale feature learning for occlusion boundary detection.

Findings

Achieved state-of-the-art F-score of 0.702 on PIOD dataset.

Improved detection speed to 0.037 seconds per image.

Effectively addressed class imbalance with a novel loss function.

Abstract

Object occlusion boundary detection is a fundamental and crucial research problem in computer vision. This is challenging to solve as encountering the extreme boundary/non-boundary class imbalance during training an object occlusion boundary detector. In this paper, we propose to address this class imbalance by up-weighting the loss contribution of false negative and false positive examples with our novel Attention Loss function. We also propose a unified end-to-end multi-task deep object occlusion boundary detection network (DOOBNet) by sharing convolutional features to simultaneously predict object boundary and occlusion orientation. DOOBNet adopts an encoder-decoder structure with skip connection in order to automatically learn multi-scale and multi-level features. We significantly surpass the state-of-the-art on the PIOD dataset (ODS F-score of .702) and the BSDS ownership dataset (ODS F-score of .555), as well as improving the detecting speed to as 0.037s per image on the PIOD dataset.