EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Scenario-Wise Comparative Evaluation of Lightweight YOLO Models for Flame and Smoke Detection

Sugi Choi, SaengGon Lee, Hyunjeong Bang, YoungJoo Song, Haiyoung Jung

Fire · 10.11.2026 · 9 (11) · 500

DOI: 10.3390/fire9110500 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Fire, online erschienen am 10.11.2026. Verfasst von Sugi Choi und 4 weiteren Autor:innen. Die Arbeit nennt 48 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Ministry of Education.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Flames and smoke vary in appearance across fire environments, limiting the ability of aggregate detection metrics to characterize model applicability. This study compared lightweight models from ten YOLO versions using an integrated dataset comprising five fire scenarios. Basic image statistics and DINOv2 feature embeddings characterized the scenario groups, while class- and scenario-specific evaluation examined performance differences obscured by aggregate metrics. All models used the same fixed data splits. Aggregate test results were summarized as means and sample standard deviations across three training runs using seeds 0, 21, and 42. The evaluation covered detection accuracy, object-size-specific average precision, inference speed, and fixed-threshold error characteristics. No single model achieved the highest class-specific AP50-95 across all scenarios. Wildland and outdoor fires yielded AP50-95 values of 0.285–0.358 for flame and 0.293–0.392 for smoke, whereas smoke AP50-95 exceeded 0.90 for all evaluated models in electrical equipment and industrial facility fires and in road and vehicle fires. Small objects also remained challenging across models. These findings demonstrate how scenario-stratified evaluation reveals detection weaknesses beyond aggregate rankings and supports baseline selection according to the target environment, object class, and computational requirements.

Förderung

  • Ministry of Education

Zitieren

APA

Choi, S., Lee, S., Bang, H., Song, Y., & Jung, H. (2026). Scenario-Wise Comparative Evaluation of Lightweight YOLO Models for Flame and Smoke Detection. Fire, 9(11), 500. https://doi.org/10.3390/fire9110500

Vancouver

Choi S, Lee S, Bang H, Song Y, Jung H. Scenario-Wise Comparative Evaluation of Lightweight YOLO Models for Flame and Smoke Detection. Fire. 2026;9(11):500. doi:10.3390/fire9110500

BibTeX
@article{Choi2026,
  author = {Choi, Sugi and Lee, SaengGon and Bang, Hyunjeong and Song, YoungJoo and Jung, Haiyoung},
  title = {Scenario-Wise Comparative Evaluation of Lightweight YOLO Models for Flame and Smoke Detection},
  journal = {Fire},
  year = {2026},
  volume = {9},
  number = {11},
  pages = {500},
  doi = {10.3390/fire9110500}
}
Kostenlos starten →