Multimodal generative AI for automated pavement condition assessment: Benchmarking model performance
PLOS One · 12.02.2026 · 21 (2) · e0340380
DOI: 10.1371/journal.pone.0340380 ↗Kurz gefasst
Fachartikel in der Zeitschrift PLOS One, online erschienen am 12.02.2026. Verfasst von Chang Xu und 3 weiteren Autor:innen. Beteiligte Einrichtung: School of Planning, Design, and Construction, Michigan State University, East Lansing, Michigan, United States of America. Hauptthemen laut MEDLINE: Benchmarking. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 87 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Great Lakes Northern Forest Cooperative Ecosystem Studies Unit.
Abstract (Originalsprache)
Accurate and efficient pavement condition assessment is essential for maintaining roadway safety and optimizing maintenance investments. However, conventional assessment methods such as manual visual inspections and specialized sensing equipment are often time-consuming, expensive, and difficult to scale across large networks. Recent advancements in generative artificial intelligence (GAI) have introduced new opportunities for automating visual interpretation tasks using street-level imagery. This study evaluates the performance of seven multimodal large language models (MLLMs) for road surface condition assessment, including three proprietary models (Gemini 2.5 Pro, OpenAI o1, and GPT-4o) and four open-source models (Gemma 3, Llama 3.2, LLaVA v1.6 Mistral, and LLaVA v1.6 Vicuna). The models were tested across four task categories relevant to pavement management: distress and feature identification, spatial pattern recognition, severity evaluation, and maintenance interval estimation. Model performance was assessed across five dimensions: response rate, response correctness, consistency, multimodal errors, and overall computational intensity and cost. Results indicate that MLLMs can interpret street-level imagery and generate task-relevant outputs in a cost-effective manner. Among the evaluated models, we recommend GPT-4o as the preferred option, as it balances responsiveness, accuracy, and computational cost.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- School of Planning, Design, and Construction, Michigan State University, East Lansing, Michigan, United States of America.
- Department of Computer Science and Engineering, Michigan State University, East Lansing, Michigan, United States of America.
Förderung
- Great Lakes Northern Forest Cooperative Ecosystem Studies Unit
Zitieren
APA
Xu, C., Shu, L., Dao, A., & Cui, Y. (2026). Multimodal generative AI for automated pavement condition assessment: Benchmarking model performance. PLOS One, 21(2), e0340380. https://doi.org/10.1371/journal.pone.0340380
Vancouver
Xu C, Shu L, Dao A, Cui Y. Multimodal generative AI for automated pavement condition assessment: Benchmarking model performance. PLOS One. 2026;21(2):e0340380. doi:10.1371/journal.pone.0340380
BibTeX
@article{Xu2026,
author = {Xu, Chang and Shu, Lei and Dao, Anh and Cui, Yue},
title = {Multimodal generative AI for automated pavement condition assessment: Benchmarking model performance},
journal = {PLOS One},
year = {2026},
volume = {21},
number = {2},
pages = {e0340380},
doi = {10.1371/journal.pone.0340380}
}
EUVIMED