Scholarlink

Effective and timely response to emergencies, such as interior fire hazards is paramount to protecting lives and property. In such a scenario, visibility is often severely reduced leading to an unclear and hazy view, hampering the efforts of individuals and increasing risks. To address this critical challenge, several De-Hazing/De-smoking techniques are already developed to make an intelligent real-time video processing system capable of automatically detecting and De-fuming hazy and smoky environments in video streams captured in fire situations. To address the visibility issues, the authors propose a new model to provide support to rescue teams and even a person enabling them to make informed decisions, locate and identify hazards, and increase the safety and efficiency of an operation. The proposed technology is a combination of deep learning and mathematical modeling of atmospheric dispersion running on CUDA parallel processing. This shows high performance as measured by metrics such as PSNR (Peak Signal-to-Noise Ratio) with a value of 36.39, SSIM (Structural Similarity Index) with a score of 0.9886. This innovative solution can be seamlessly integrated with unmanned technology to improve mobility and map more convenient scenarios in crowded areas. Contribute significantly to the success and efficiency of operations by solving critical challenges that arise in emergency situations.

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

From Haze and Smoke to Clarity: An Integration of Deep Learning and Atmospheric Model

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Tashneet Kaur,
  • Dhruv Kumar,
  • Sita Rani,
  • Kiran Jyoti

摘要

Effective and timely response to emergencies, such as interior fire hazards is paramount to protecting lives and property. In such a scenario, visibility is often severely reduced leading to an unclear and hazy view, hampering the efforts of individuals and increasing risks. To address this critical challenge, several De-Hazing/De-smoking techniques are already developed to make an intelligent real-time video processing system capable of automatically detecting and De-fuming hazy and smoky environments in video streams captured in fire situations. To address the visibility issues, the authors propose a new model to provide support to rescue teams and even a person enabling them to make informed decisions, locate and identify hazards, and increase the safety and efficiency of an operation. The proposed technology is a combination of deep learning and mathematical modeling of atmospheric dispersion running on CUDA parallel processing. This shows high performance as measured by metrics such as PSNR (Peak Signal-to-Noise Ratio) with a value of 36.39, SSIM (Structural Similarity Index) with a score of 0.9886. This innovative solution can be seamlessly integrated with unmanned technology to improve mobility and map more convenient scenarios in crowded areas. Contribute significantly to the success and efficiency of operations by solving critical challenges that arise in emergency situations.