无人机视角下的火灾烟雾检测模型

近年来火灾对人类社会造成的危害性成逐年上涨的趋势，仅2023年上半年因火灾造成的直接经济损失就高达近40亿元。对于火灾的常预防、早发现是减少火灾危害的主要需求，而火灾烟雾的实时检测技术正是对标该需求的一项重要安全技术。本团队面向上述实际问题，提出了无人机视角下的火灾烟雾检测模型，能够实现基于无人机视角对火灾和烟雾进行实时、精确的检测，并且算法具备较强的环境适应能力，能够应对天气恶劣、画质模糊，干扰物多等情况对检测效果带来的挑战，进而在灾难抢险、环境监测等领域助力完成及时预警和实时搜救等任务。

In recent years, the harm inflicted by fires on human society has been on a steady upward trend. In the first half of 2023 alone, the direct economic losses caused by fires reached nearly 4 billion yuan. Regular fire prevention and early detection are the core requirements for mitigating fire hazards, and real-time fire smoke detection technology is a critical safety technology that addresses this demand. To solve the above-mentioned practical problems, our team proposes a fire and smoke detection model based on the UAV perspective. This model enables real-time and accurate detection of fires and smoke from UAV-captured footage, and the algorithm exhibits strong environmental adaptability, capable of overcoming challenges to detection performance arising from harsh weather, blurry image quality, abundant interfering objects and other scenarios. Consequently, it can support timely early warning and real-time search and rescue tasks in fields such as disaster rescue and relief, as well as environmental monitoring.