【摘要】 车道检测是自动驾驶不可缺少的一部分，但目前车道检测算法在同时保证高准确率和快检测速度方面还有待进一步提高。在LaneATT上改进得到了新的车道检测算法——LaneEcaATT。通过引入了通道注意力机制ECAnet对特征图进行处理得到局部特征，然后与全局特征相结合进行车道检测。在两个公开数据集上的实验结果表明，改进后的算法LaneEcaATT在Tusimple数据集上优于对比算法，在CULane数据集上采用Resnet18作为主干网时也优于对比算法，较好地兼顾了高准确率和快检测速度。更多还原

【Abstract】 Lane detection is an indispensable part of automatic driving, but the current lane detection algorithm needs to be further improved in terms of ensuring high accuracy and fast detection speed. A new lane detection algorithm LaneEcaATT is improved based on LaneATT. By introducing the channel attention mechanism ECAnet, the local features are obtained by processing the feature map, and then combined with the global features for lane detection. The experimental results on two public datasets show that the improved algorithm LaneEcaATT is better than the comparison algorithm on the Tusimple dataset, and is also better than the comparison algorithm when Resnet18 is used as the backbone network on the CULane dataset, which gives better consideration to high accuracy and fast detection speed.更多还原