FireDet: A framework for precise detection and localization of small-scale fire

FireDet: A framework for precise detection and localization of small-scale fire

Overview

FireDet integrates small-flame detection with calibrated binocular localization to support precise early warning for small-scale fire sources.

Core Contributions

• Develops the FSDNet flame detector for challenging small-scale fire targets.
• Builds an integrated FireDet framework for detection, binocular localization, and parameter optimization.
• Provides dedicated datasets for comprehensive flame detection, small-target generalization, stereo localization, and calibration.

Publication Information

• Authors: Jinyang Dai, Qiuyang Cao, Xiaoning Song, Bence Pan, Qixing Zhang
• Journal: Fire Safety Journal
• Publication date: 2026-07-01
• DOI: 10.1016/j.firesaf.2026.104737
• Publisher: Elsevier BV
• Status: Published

Citation

Jinyang Dai, Qiuyang Cao, Xiaoning Song, Bence Pan, Qixing Zhang. “FireDet: A framework for precise detection and localization of small-scale fire.” Fire Safety Journal, vol. 162, 104737 (2026). https://doi.org/10.1016/j.firesaf.2026.104737

Keywords

small-scale fire, flame detection, binocular localization, early warning, computer vision

Links

• DOI
• Publisher page
• Code or data repository