基于改进的YOLOv7自动驾驶目标识别算法

Improved YOLOv7 Automatic Driving Object Recognition Algorithm

为了解决自动驾驶场景下车辆目标小且具有低分辨率、模糊背景和少量的特征信息等问题,本文提出了一种基于融合CBAM注意力机制的改进YOLOv7车辆目标识别方法。首先,在传统的YOLOv7目标检测算法的基础上,替换2020的检测尺度为160160,增加浅层特征提取能力,其次引入了CBAM注意力机制来增强模型对周围车辆环境的感知能力和定位精度,最后引入NWD度量,改进原CIoU损失函数对小目标位置偏差敏感的问题,利用Wasserstein距离度量高斯分布的包围框的相似度,来增加对较小目标的检测。结果表明,改进后模型在FPS基本保持不变检测精确度(P)和平均检测精度(mAP)方面分别提高了1.5%和3.2%。

To address the challenges of small and low-resolution vehicle targets,blurry backgrounds,and limited distinctive features in autonomous driving scenarios,an improved YOLOv7 vehicle detection method based on the integrated CBAM attention mechanism is proposed Firstly,on the basis of the traditional YOLOv7 object detection algorithm,the detection scale of 20×20 is replaced with 160×160 to increase the ability of shallow feature extraction,secondly,the CBAM attention mechanism is introduced to enhance the model's perception and positioning accuracy of small targets,and finally the NWD metric is introduced to improve the sensitivity of the original CIoU loss function to the position deviation of small targets,and the similarity of the bounding box of Gaussian distribution is measured by using the Wasserstein distance to increase the detection of small targets.The results demonstrate that the improved model maintains a similar Frames Per Second(FPS)while achieving a 1.5%increase in precision(P)and a 3.2%increase in average precision(mAP)for object detection.