Generative Learning for Post-Processing Semantic Segmentation Predictions: A Lightweight Conditional Generative Adversarial Network based on Pix2pix to Improve the Extraction of Road Surface Areas

Cira, Calimanut-ionut and Manso Callejo, Miguel Angel and Alcarria Garrido, Ramon Pablo and Fernandez Pareja, Maria Teresa and Bordel Sanchez, Borja and Serradilla Garcia, Francisco (2021). Generative Learning for Post-Processing Semantic Segmentation Predictions: A Lightweight Conditional Generative Adversarial Network based on Pix2pix to Improve the Extraction of Road Surface Areas. "Deep Learning Algorithms for Land Use Change Detection", v. 10 (n. 1); pp. 1-15. ISSN 2073-445X. https://doi.org/10.3390/land10010079.

Description

Title: Generative Learning for Post-Processing Semantic Segmentation Predictions: A Lightweight Conditional Generative Adversarial Network based on Pix2pix to Improve the Extraction of Road Surface Areas
Author/s:
  • Cira, Calimanut-ionut
  • Manso Callejo, Miguel Angel
  • Alcarria Garrido, Ramon Pablo
  • Fernandez Pareja, Maria Teresa
  • Bordel Sanchez, Borja
  • Serradilla Garcia, Francisco
Item Type: Article
Título de Revista/Publicación: Deep Learning Algorithms for Land Use Change Detection
Date: 16 January 2021
ISSN: 2073-445X
Volume: 10
Subjects:
Freetext Keywords: Conditional Generative Adversarial Network; Generative learning; Postprocessing semantic segmentation predictions; Road extraction; Road surface areas
Faculty: E.T.S.I. en Topografía, Geodesia y Cartografía (UPM)
Department: Ingeniería Cartográfica y Topografía
Creative Commons Licenses: Recognition - No derivative works - Non commercial

Full text

[img]
Preview
PDF - Requires a PDF viewer, such as GSview, Xpdf or Adobe Acrobat Reader
Download (2MB) | Preview

Abstract

Remote sensing experts have been actively using deep neural networks to solve extraction tasks in high-resolution aerial imagery by means of supervised semantic segmentation operations. However, the extraction operation is imperfect, due to the complex nature of geospatial objects, limitations of sensing resolution, or occlusions present in the scenes. In this work, we tackle the challenge of postprocessing semantic segmentation predictions of road surface areas obtained with a state-of-the-art segmentation model and present a technique based on generative learning and image-to-image translations concepts to improve these initial segmentation predictions. The proposed model is a conditional Generative Adversarial Network based on Pix2pix, heavily modified for computational efficiency (92.4% decrease in the number of parameters in the generator network and 61.3% decrease in the discriminator network). The model is trained to learn the distribution of the road network present in official cartography, using a novel dataset containing 6784 tiles of 256 × 256 pixels in size, covering representative areas of Spain. Afterwards, we conduct a metrical comparison using the Intersection over Union (IoU) score (measuring the ratio between the overlap and union areas) on a novel testing set containing 1696 tiles (unseen during training) and observe a maximum increase of 11.6% in the IoU score (from 0.6726 to 0.7515). In the end, we conduct a qualitative comparison to visually assess the effectiveness of the technique and observe great improvements with respect to the initial semantic segmentation predictions.

More information

Item ID: 65915
DC Identifier: http://oa.upm.es/65915/
OAI Identifier: oai:oa.upm.es:65915
DOI: 10.3390/land10010079
Official URL: https://www.mdpi.com/2073-445X/10/1/79
Deposited by: Memoria Investigacion
Deposited on: 08 Feb 2021 17:45
Last Modified: 10 Feb 2021 14:57
  • Logo InvestigaM (UPM)
  • Logo GEOUP4
  • Logo Open Access
  • Open Access
  • Logo Sherpa/Romeo
    Check whether the anglo-saxon journal in which you have published an article allows you to also publish it under open access.
  • Logo Dulcinea
    Check whether the spanish journal in which you have published an article allows you to also publish it under open access.
  • Logo de Recolecta
  • Logo del Observatorio I+D+i UPM
  • Logo de OpenCourseWare UPM