摘要
为提高船舶火灾检测的响应速度,针对海洋环境下船舶火灾的特点,提出一种基于图像增强算法和YOLO算法的海上船舶火灾检测方法.构建一套面向海上船舶火灾的数据集,对数据集图像进行增强处理,以突出火焰细节.对网络的头部结构进行改进,只保留其中的特征金字塔网络(feature pyramid network,FPN)结构,在保证模型精度的前提下减少模型参数,加快计算速度.通过实验进行验证,结果表明所提出方法的检测精确度达到97.2％,平均检测时间缩短至6.9 ms.与改进前的YOLOv5s相比,其检测精度提高了,检测速度提升了 22.5％.所提出方法能对火焰进行实时监测,识别准确度高,检测速度快,能提供一种在海洋环境下有效的船舶火灾检测及救援技术方案.
Abstract
In order to improve the response speed of ship fire detection,according to the characteristics of ship fire in maritime environment,a maritime ship fire detection method based on the image enhancement algorithm and the YOLO algorithm is proposed.A set of data set for maritime ship fire is constructed,and the images of the data set are enhanced to highlight the flame details.The head structure of the network is improved,and only the feature pyramid network(FPN)is retained.On the premise of ensuring the precision of the model,the model parameters are reduced and the calculation speed is accelerated.The experimental verification is carried out.The results show that,the detection accuracy of the proposed method is increased to 97.2％,and the average detection time is reduced to 6.9 ms.Compared with the original YOLOv5s,the detection accuracy of the proposed method is increased,and its detection speed is increased by 22.5％.The proposed method can monitor the flame in real time with high identification accuracy and fast detection speed.It can provide an effective technical scheme for ship fire detection and rescue in maritime environment.
基金项目
上海市"科技创新行动计划"社会发展科技攻关项目(21DZ1205803)
NETL
NSTL
万方数据