基于目标特征蒸馏的车道线检测

Lane Detection Based on Object Feature Distillation

为满足自动驾驶系统实时性要求,现有方法将编码器输出的特征图直接上采样进行像素级预测,从而忽略了解码器对于细节特征预测的重要性。为解决该问题,提出一种通用的基于目标特征蒸馏的车道线检测框架。首先,在使用直接上采样方式的网络中,增加一个具有较强特征预测能力的解码器;然后,在网络训练阶段,通过知识蒸馏技术将解码器生成的预测结果作为软目标,以使直接上采样分支学习到更为详尽的车道线信息,让其具有解码器的较强特征预测能力;最后,在网络推理阶段仅需使用直接上采样分支,而无需对解码器进行前向计算,因此相比现有模型在不增加额外计算成本的同时还能提高车道线检测性能。为验证本框架的有效性,将其应用到诸如SCNN、Deeplabv1、Res Net等多种主流的车道线分割方法上。实验结果表明:在不增加额外复杂度的条件下,所提出方法在Culane数据集上获得了更高的F1-Measure评分。

In order to meet the real-time requirements of the autonomous driving system,the existing method directly upsampling the encoder’s output feature map to pixel-wise prediction,thus neglecting the importance of the decoder for the prediction of detail features.In order to solve this problem,this paper proposes a general lane detection framework based on object feature distillation.Firstly,a decoder with strong feature prediction ability is added to the network using directly upsampling method;Then,in the network training stage,the prediction results generated by the decoder are regarded as soft targets through knowledge distillation technology,so that the directly upsampling branch can learn more detailed lane information and have strong feature prediction ability of the decoder;Finally,in the stage of network inference,we only need to use the directly upsampling branch instead of the forward calculation of the decoder,so compared with the existing model,it can improve the lane detection performance without additional cost.In order to verify the effectiveness of this framework,it is applied to many mainstream lane segmentation methods such as SCNN,Deeplabv1,Res Net and etc.The experiment shows that under the condition of no additional complexity,compared with the above methods,the F1 score of Culane dataset is higher.