Deep Fence Estimation using Stereo Guidance and Adversarial Learning

TL;DR

This paper introduces a novel deep learning approach for fence removal in images, utilizing stereo guidance masks and a directional connectivity loss to improve segmentation accuracy of wire fences.

Contribution

It proposes a new fence guidance mask derived from stereo images and a directional connectivity loss, enhancing fence segmentation performance over existing methods.

Findings

Outperforms state-of-the-art fence removal techniques

Effective in detecting thin wires and complex fence structures

Demonstrated on real-world scenarios with superior results

Abstract

People capture memorable images of events and exhibits that are often occluded by a wire mesh loosely termed as fence. Recent works in removing fence have limited performance due to the difficulty in initial fence segmentation. This work aims to accurately segment fence using a novel fence guidance mask (FM) generated from stereo image pair. This binary guidance mask contains deterministic cues about the structure of fence and is given as additional input to the deep fence estimation model. We also introduce a directional connectivity loss (DCL), which is used alongside adversarial loss to precisely detect thin wires. Experimental results obtained on real world scenarios demonstrate the superiority of proposed method over state-of-the-art techniques.