PLGAN: Generative Adversarial Networks for Power-Line Segmentation in Aerial Images

TL;DR

PLGAN is a novel GAN-based method that improves power-line segmentation in aerial images by embedding GAN features into a segmentation network and using a new loss function in Hough space, outperforming previous methods.

Contribution

Introduces PLGAN, a GAN-based approach that enhances power-line segmentation by integrating GAN features and a Hough space loss function for better accuracy.

Findings

PLGAN outperforms state-of-the-art segmentation methods.

The Hough space loss improves thin line detection.

Embedding GAN features enhances segmentation quality.

Abstract

Accurate segmentation of power lines in various aerial images is very important for UAV flight safety. The complex background and very thin structures of power lines, however, make it an inherently difficult task in computer vision. This paper presents PLGAN, a simple yet effective method based on generative adversarial networks, to segment power lines from aerial images with different backgrounds. Instead of directly using the adversarial networks to generate the segmentation, we take their certain decoding features and embed them into another semantic segmentation network by considering more context, geometry, and appearance information of power lines. We further exploit the appropriate form of the generated images for high-quality feature embedding and define a new loss function in the Hough-transform parameter space to enhance the segmentation of very thin power lines. Extensive experiments and comprehensive analysis demonstrate that our proposed PLGAN outperforms the prior state-of-the-art methods for semantic segmentation and line detection.