Scholarlink

In fire emergencies, swift and adaptive response systems are critical to saving lives and minimizing damage. Yet, creating such systems involves tackling significant challenges: reducing processing delays, ensuring reliability across varied environments, and enabling scalability for complex layouts. Real-time video analysis must quickly and accurately detect fires, adapting to diverse lighting, building structures, and potential visual obstructions. Additionally, dynamic route optimization must seamlessly adjust to the fire’s spread, guiding occupants along the safest, least crowded paths. Our proposed solution rises to these challenges by combining cutting-edge deep learning with real-time video processing, deploying the precision of the Inception V3 CNN model for fire detection and the adaptability of Dijkstra’s algorithm for route planning. Capable of handling multi-camera feeds across expansive layouts, this system is both scalable and highly responsive. Real-time guidance, broadcast through a public address system, enables prompt, coordinated evacuations, setting a new benchmark in intelligent fire safety technology.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Intelligent Fire Detection and Dynamic Evacuation System Using Deep Learning and Real-Time Video Analysis

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • D. Roja Ramani,
  • V. Sandhya Rani,
  • Ganesh Dagadi,
  • Harshavardhan V. Mantri,
  • S. Preetham,
  • Nipun Goyal

摘要

In fire emergencies, swift and adaptive response systems are critical to saving lives and minimizing damage. Yet, creating such systems involves tackling significant challenges: reducing processing delays, ensuring reliability across varied environments, and enabling scalability for complex layouts. Real-time video analysis must quickly and accurately detect fires, adapting to diverse lighting, building structures, and potential visual obstructions. Additionally, dynamic route optimization must seamlessly adjust to the fire’s spread, guiding occupants along the safest, least crowded paths. Our proposed solution rises to these challenges by combining cutting-edge deep learning with real-time video processing, deploying the precision of the Inception V3 CNN model for fire detection and the adaptability of Dijkstra’s algorithm for route planning. Capable of handling multi-camera feeds across expansive layouts, this system is both scalable and highly responsive. Real-time guidance, broadcast through a public address system, enables prompt, coordinated evacuations, setting a new benchmark in intelligent fire safety technology.