Enhanced Lithium-Ion Battery Fire Detection Using Novel Algorithm YFSNet

Strategic Insights -

Researchers at the Civil Aviation Flight University of China have developed an advanced algorithm, YFSNet, to improve lithium-ion battery fire detection in complex backgrounds. The novel algorithm integrates feature refinement and streamlined architecture, significantly enhancing detection precision and real-time performance. YFSNet was trained on a dataset of 2300 high-quality images, achieving a detection precision of 99.6% and a marked improvement in inference speed, from 49.75 FPS to 116.28 FPS.

Key Takeaways:

  • YFSNet is a novel algorithm that integrates advanced modules for richer feature extraction and streamlined architecture, enhancing detection precision and real-time performance.
  • The algorithm was trained on a dataset of 2300 high-quality images and demonstrated a detection precision of 99.6%, compared to 95.6% with the traditional YOLOv8n model.
  • The inference speed of YFSNet showed a marked improvement, increasing from 49.75 FPS to 116.28 FPS.
  • Although the recall rate dropped from 97.7% to 93.1%, the overall performance remained robust, with a high F1-score and detection accuracy.
  • The YFSNet algorithm has the potential for reliable and efficient battery fire detection in fire safety systems.
  • Li Deng, Quanyi Liu, and Di Kang are the authors of the research paper, published in the journal Fire (MDPI).
  • The study was funded by the National Natural Science Foundation of China, Key Laboratory Project of Sichuan Province, Fundamental Research Funds For The Central Universities, Aviation Science Fund, and Quanyi Liu.

Statistics:

  • Detection precision: 99.6% (YFSNet) vs. 95.6% (traditional YOLOv8n model)
  • Inference speed: 116.28 FPS (YFSNet) vs. 49.75 FPS (traditional YOLOv8n model)
  • Recall rate: 97.7% (traditional YOLOv8n model) vs. 93.1% (YFSNet)
  • F1-score and detection accuracy: robust and reliable

Sources:

  • An Improved Lithium-Ion Battery Fire and Smoke Detection Method Based on the YOLOv8 Algorithm. Fire, 2025,8(6):214. (Fire - https://www.mdpi.com/journal/fire)
  • NewsRx. Researchers from Civil Aviation Flight University of China Report on Findings in Physics (An Improved Lithium-Ion Battery Fire and Smoke Detection Method Based on the YOLOv8 Algorithm). Physics Week. July 8, 2025; p 672.