A feasibility study of applying generative deep learning models for map labeling

ABSTRACT

The automation of map labeling is an ongoing research challenge. Currently, the map labeling algorithms are based on rules defined by experts for optimizing the placement of the text labels on maps. In this paper, we investigate the feasibility of using well-labeled map samples as a source of knowledge for automating the labeling process. The basic idea is to train deep learning models, specifically the generative models CycleGAN and Pix2Pix, on a large number of map examples. Then, the trained models are used to predict good locations of the labels given unlabeled raster maps. We compare the results obtained by the deep learning models to manual map labeling and a state-of-the-art optimization-based labeling method. A quantitative evaluation is performed in terms of legibility, association and map readability as well as a visual evaluation performed by three professional cartographers. The evaluation indicates that the deep learning models are capable of finding appropriate positions for the labels, but that they, in this implementation, are not well suited for selecting the labels to show and to determine the size of the labels. The result provides valuable insights into the current capabilities of generative models for such task, while also identifying the key challenges that will shape future research directions.

KEYWORDS:

• Map labeling
• automated cartography
• machine learning
• deep learning
• image synthesis
• generative adversarial networks

Acknowledgments

Thanks to Transport of London and T-Kartor for providing the map data. We are grateful to the tree cartographic experts who conducted the visual evaluation: Micael Runnström (Lund University), Mikael Johansson (Lantmäteriet – the National Mapping Agency in Sweden) and Åsa Nilsson (T-Kartor). Also, thanks to Andreas Oxenstierna, Kai-Florian Richter and Kalle Åström for advice.

The computations were enabled by resources provided by the Swedish National Infrastructure for Computing (SNIC) at Chalmers Centre for Computational Science and Engineering (C3SE), partially funded by the Swedish Research Council through grant agreement no. 2018-05973. Project agreement SNIC 2022/22-426.

Thanks to the QGIS project for providing the optimisation labeling and other tools used in the study.

Thanks to the anonymous reviewers for constructive comments on early versions of this paper.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The map data used in this study is commercial and under a license form that does not allow it to be distributed.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15230406.2023.2291051.

Notes

1. https://github.com/Rachidovij/AutomaticMapLabeling

2. https://www.citywayfinding.com

3. Transport for London (2009) Street map design standard – Issue 1.

Transport for London (2011) Street map design standard – Issue 2.

4. https://www.t-kartor.com/

5. https://desktop.arcgis.com/en/arcmap/10.3/map/working-with-text/what-is-maplex-.htm

6. Transport for London (2009) Street map design standard – Issue 1.

Transport for London (2011) Street map design standard – Issue 2.

7. https://rasterio.readthedocs.io/en/latest/

8. https://qgis.org/en/site/

9. http://pal.heig-vd.ch/

10. https://github.com/Rachidovij/AutomaticMapLabeling

Additional information

Funding

This study was funded by eSSENCE@LU 7:1: “Data-driven automation of map labeling – enabling affordable high-quality maps” as part of the strategic research area eSSENCE (financed by the Swedish research council). Financial support has also been provided by Lund University.