Learning Region Similarities via Graph-Based Deep Metric Learning

Zhao, Yunxiang; Qi, Jianzhong; Trisedya, Bayu Distiawan; Su, Yixin; Zhang, Rui; Ren, Hongguang

doi:10.1109/tkde.2023.3253802

Cited by 4 publications

(1 citation statement)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Meanwhile, the images from the other cities are more or less correctly evaluated by the GED, receiving a high value and thus a poor graph matching, as verified by the other graph evaluating metrics. This issue is in line with observations of Hu et al (2020) and Zhao et al (2023) who adapted the GED themselves to enable its use for their specific research question. We assume that this discrepancy in the GED, particularly for Paris, appears due to the low number of nodes and edges contained in the relatively rural road network depicted in the images of Paris.…”

Section: Discussionsupporting

confidence: 73%

Road Network Mapping from Multispectral Satellite Imagery: Leveraging Deep Learning and Spectral Bands

Hollendonner,

Alinaghi,

Giannopoulos

2024

AGILE GIScience Ser.

View full text Add to dashboard Cite

Abstract. Updating road networks in rapidly changing urban landscapes is an important but difficult task, often challenged by the complexity and errors of manual mapping processes. Traditional methods that primarily use RGB satellite imagery struggle with obstacles in the environment and varying road structures, leading to limitations in global data processing. This paper presents an innovative approach that utilizes deep learning and multispectral satellite imagery to improve road network extraction and mapping. By exploring U-Net models with DenseNet backbones and integrating different spectral bands we apply semantic segmentation and extensive post-processing techniques to create georeferenced road networks. We trained two identical models to evaluate the impact of using images created from specially selected multispectral bands rather than conventional RGB images. Our experiments demonstrate the positive impact of using multispectral bands, by improving the results of the metrics Intersection over Union (IoU) by 6.5%, F1 by 5.4%, and the newly proposed relative graph edit distance (relGED) and topology metrics by 2.2% and 2.6% respectively.

show abstract

Section: Discussionsupporting

confidence: 73%