摘要
车辆检测是智能交通、无人驾驶等系统得以实现的重要支撑性技术。低精度或低速度的车辆检测器应用受限,因此提出了一种快速准确的车辆检测器。首先,前端特征提取网络VGG16由MobileNetV3_Large替代,减少了参数量和计算量,并增加了对高维特征的提取能力;其次,利用特征金字塔思想构建双向加权融合网络,有效融合不同尺度的特征,获取多维度的车辆特征;最后在特征提取层引入高效通道注意力,重新标定不同特征通道的重要性,进一步提高模型性能。与SSD相比,所提出的模型在KITTI数据集和BDD 100 K数据集上分别将平均精度提高了7.50%和3.50%,并具有实时检测能力(超过40 fps),在检测精度和速度方面有更好的平衡,说明了方法的有效性。
Vehicle detection is an important supporting technology for the realization of intelligent transportation,autonomous driving,etc.Poor accuracy or low inference vehicle detectors are limited in application,therefore this paper proposes a fast and accurate vehicle detector.First,the front-end feature extraction network VGG16 is replaced by MobileNetV3_Large,which reduces the number of parameters and computation,and increases the ability to extract high-dimensional features.Next,the feature pyramid idea is used to construct a weighted bi-directional fusion network to obtain multi-dimensional vehicle features;In the end,introducing efficient channel attention in the feature extraction layer to re-calibrate the importance of different feature channels and further improve the model performance.Compared with SSD,our proposed model improves mAP by 7.50%and 3.50%on KITTI dataset and BDD 100 K dataset,and with real-time inference(more than 40 fps),it reports a better trade-off in terms of detection accuracy and speed,illustrating the effectiveness of our method.
作者
张奇
陈梦蝶
赵杰
Zhang Qi;Chen Mengdie;Zhao Jie(School of Electronics and Information,Xi'an Polytechnic University,Xi'an 710048,China)
出处
《国外电子测量技术》
北大核心
2023年第1期41-48,共8页
Foreign Electronic Measurement Technology
基金
西安市碑林区应用技术研发项目(GX2007)资助
