Work

Deep Pavements Framework: Combining AI Tools And Collaborative Terrestrial Imagery For Pathway Mapping

Year: 2024

Type: conference-paper

Abstract: Pedestrian mobility in urban environments is crucial for achieving UN SDGs, and mapping is essential for long-term planning. Open street-level imagery, like Mapillary, presents opportunities and challenges for data extraction. The project aims to study pathways, verify their existence, categorize them, and identify surface materials. The Deep Pavements framework proposes modular solutions for these challenges, including dataset labeling, data processing, sample generation, and user guidance. The implementation involves AI algorithms for data extraction, emphasizing hallucination testing to improve semantic understanding. The framework is open, reproducible, and standards-based, with ongoing development and plans for image filtering and integrating additional pavement traits.

Source: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.14250770

Authors: de Moraes Vestena, Kauê, Camboim, Silvana, dos Santos, Daniel Rodrigues

Institutions: Universidade Federal do Paraná, Military Institute of Engineering

Language: English

Topic: Automated Road and Building Extraction

Subfield: Ocean Engineering

Field: Engineering

Domain: Physical Sciences

SDGs: Sustainable cities and communities, Industry, innovation and infrastructure

Open Access status: green

OpenAlex ID: W6911762485

FWCI:

Cites:

Cited by:

Related to: