基于AOD-Net和SSD的雾天车辆和行人检测

Vehicle and Pedestrian Detection Based on AOD-Net and SSD Algorithm in Hazy Environment

为了提高智能汽车在恶劣天气下的环境感知能力,提出一种针对雾天环境下车辆和行人的检测方法。将AOD-Net去雾算法与SSD目标检测算法相结合,实现了城市交通雾天环境下的车辆和行人检测。利用去雾处理后的雾天图片和原始带雾图片分别进行目标检测模型训练,并在不同雾浓度等级的交通环境下进行车辆和行人检测,结果显示:AOD-Net与SSD网络相结合得到的检测mAP值可达75.8%,比SSD算法的mAP值高4.1%,表明AOD-Net与SSD网络相结合的算法能更加有效地检测带雾图片中的车辆和行人。

In order to improve the environment perception ability of intelligent vehicle in severe weather,a novel method for vehicle and pedestrian detection in hazy environment is proposed.Combining the AOD-Net dehazing algorithm with the SSD target detection technology,pedestrians and vehicles can be detected in the hazy urban traffic environment.The dehazing and hazy images are used for target detection model training,and the models are tested to detect vehicle and pedestrian in traffic environments with different fog density levels.The results show that the mAP of the combination of AOD-Net and SSD network algorithm could reach 75.8%,which is 4.1%higher than that of the SSD algorithm alone.It indicate that the algorithm combining AOD-Net and SSD network can detect vehicles and pedestrians more effectively in hazy images.