This work develops an omni-deterrence system to resolve the bird strike problem in airport. Through AI models, edge computing, dual telephoto lenses, and motor control, it can automatically identify and dynamically track birds, then plays sound to repel birds. The system integrates LLMs and RAG technologies to quickly obtain bird information and prevent risks. This design achieves automatic bird deterrence to reduce manpower requirements and decrease bird strike-related flight safety incidents.

1. An automatic system for detecting, tracking, identifying, and deterring birds;
2. Optimized AI models on edge computer with mAP of over 90%;
3. Integrating computer vision and mechatronics to improve the accuracy of small object identification, dynamic tracking, and real-time control stability;
4. Automatically playing different scare sounds to deter birds;
5. Providing omni-reducing bird strike risk in taking-off and landing areas, generalizing and quickly installing in different locations.

Having been successfully implemented in four major fields:
  
  1. Airport: The system can effectively detect, track, identify, and deter birds;
  2. Recreation centers and farms: To deter Formosan macaques without harming them;
  3. Agriculture: In rice paddies and orchards, deter birds or monkeys only upon detection them;
  4. Campus: Providing online monitoring of bird ecology and information is integrated to our iOS App, EEBirds.
  5. Detection mAP: In average, higher than 90% in different applications.