基于深度学习的火焰检测算法设计  

Design of Flame Detection Algorithm Based on Deep Learning

作　　者：姚一 鞠全勇[2] 吕章锟 

机构地区：[1]盐城工学院机械工程学院,江苏 盐城 [2]金陵科技学院机电学院,江苏 南京

出　　处：《人工智能与机器人研究》2024年第2期344-353,共10页Artificial Intelligence and Robotics Research

摘　　要：为了更好地实现消防机器人对于火源的自主识别,结合YOLOv5s提出一种改进的火焰检测方法。在YOLOv5s的主干网络和检测头中添加CBAM注意力模块,能够自适应地对特征图进行调整,使得网络能够更好地关注重要的特征,从而降低火焰检测的小目标漏检率。使用C3-DBB模块替换卷积层,C3-DBB模块采用了密集的连接和多尺度的感受野,从而能够更好地捕捉目标的细节信息,减少信息损失。使用ConvNeXt V2模块替换主干网络模块,从而获得更大的感受野和更好的特征提取能力,提高目标检测的精度。实验结果表明,与YOLOv5s原始模型相比,改进后的模型平均精度提升5.8%,同时降低了模型大小。In order to better realize the autonomous recognition of fire sources by firefighting robots, an improved flame detection method is proposed in combination with YOLOv5s. Adding the CBAM attention module to the backbone network and detection head of YOLOv5s can adaptively adjust the feature map, enabling the network to better focus on important features, thus reducing the small target miss detection rate of flame detection. Replacing the convolutional layer with the C3-DBB module, the C3-DBB module employs dense connectivity and a multi-scale receptive field, which enables better capture of detailed information about the target and reduces information loss. The ConvNeXt V2 module is used to replace the backbone network module, thus obtaining a larger receptive field and better feature extraction capability to improve the accuracy of target detection. The experimental results show that the improved model improves the average accuracy by 5.8% compared to the original YOLOv5s model, while reducing the model size.

关 键 词：火焰检测 注意力机制 YOLOv5s算法 深度学习 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]