Sign in Subscribe
Aviation

Advancements in Aviation Safety: Researchers Develop Lightweight Deep Learning Framework for FOD Detection

Strategic Insights

— 1 min read

Researchers at Nanjing University of Aeronautics and Astronautics have made a breakthrough in aviation safety with the development of a lightweight deep learning framework for detecting Foreign Object Debris (FOD) in surveillance imagery. The new framework, called LiteFODNet, has been shown to achieve higher accuracy and efficiency in detecting small, complex objects on airport runways, reducing the risk of aircraft collisions.

Key Takeaways:

  • LiteFODNet is a lightweight and data-efficient deep learning framework specifically designed for intelligent FOD detection in surveillance imagery.
  • The framework consists of four novel architectural modules: Compact Multi-Scale Pooling, Spatial-Channel Reducer, Feature Focus Module, and Split Path Attention.
  • LiteFODNet achieves 0.8888% higher mAP@50-95 than YOLOv8n while reducing parameters by 16.39%, inference time by 27.77%, and GFLOPs by 3.66%.
  • The research has been peer-reviewed and has been shown to offer an intelligent, high-performance solution for real-time FOD detection under constrained resources.
  • LiteFODNet has strong potential for deployment in aviation safety monitoring systems.
  • The research team includes Ali Khan, Izhar Ahmed Khan, Hai Deng, Somaiya Khan, Mohammed A. M. Elhassan, Rizwan Khan, and Mohammed Alsuhaibani.
  • The research was financially supported by the Deanship of Graduate Studies and Scientific Research at Qassim University.
  • The authors conclude that LiteFODNet offers a reliable and efficient solution for aviation safety monitoring, reducing the risk of aircraft collisions and improving passenger safety.

Statistics:

  • 0.8888%: The improvement in mAP@50-95 achieved by LiteFODNet over YOLOv8n.
  • 16.39%: The reduction in parameters achieved by LiteFODNet.
  • 27.77%: The reduction in inference time achieved by LiteFODNet.
  • 3.66%: The reduction in GFLOPs achieved by LiteFODNet.
  • 2025: The year in which the research was conducted and published.
  • November 1, 2025: The date on which the research was published in the Journal of Transportation.

Sources:

  • [Source 1] Litefodnet: a Lightweight Deep Learning Model for Intelligent Detection of Small Objects In Runway Surveillance Data. Intelligent Data Analysis, 2025.
  • [Source 2] NewsRx LLC, News Article: Studies from Nanjing University of Aeronautics and Astronautics in the Area of Aviation Safety Described (Litefodnet: a Lightweight Deep Learning Model for Intelligent Detection of Small Objects In Runway Surveillance Data). Journal of Transportation. November 1, 2025; p 272.

Read more