基于You Only Look Once v2优化算法的车辆实时检测

Real-time Vehicle Detection Based on Optimized You Only Look Once v2 Algorithm

针对基于You Only Look Once v2算法的目标检测存在精度低及稳健性差的问题,提出一种车辆目标实时检测的You Only Look Once v2优化算法;该算法以You Only Look Once v2算法为基础,通过增加网络深度,增强特征提取能力,同时,通过添加残差模块,解决网络深度增加带来的梯度消失或弥散问题;该方法将网络结构中低层特征与高层特征进行融合,提升对小目标车辆的检测精度。结果表明,通过在KITTI数据集上进行测试,优化后的算法在检测速度不变的情况下,提高了车辆目标检测精度,平均精度达到0.94,同时提升了小目标检测的准确性。

Aiming at the problem of low precision and poor robustness of object detection based on You Only Look Once v2 algorithm,a real-time detection method for vehicle object was proposed named optimized You Only Look Once v2 algorithm.Based on You Only Look Once v2 algorithm,by increasing the depth of network,the feature extraction capability was enhanced.By adding the residual module,the gradient disappearance or dispersion problem caused by the increase of network depth was solved.The detection accuracy of small target vehicles was improved by combining low-level features and high-level features in the network structure.The results show that by testing on KITTI dataset,the optimized algorithm improves the accuracy of vehicle object detection without decreasing the detection speed.The average precision reaches 0.94,and the accuracy of small target detection is improved.