At present, a major challenge for the application of automatic driving technology is the accurate prediction of vehicle trajectory. With the vigorous development of computer technology and the emergence of convolution depth neural network, the accuracy of prediction results has been improved. But, the depth, width of the network and image resolution are still important reasons that restrict the accuracy of the model and the prediction results. The main innovation of this paper is the combination of RESNET network and efficient net network, which not only greatly increases the network depth, but also comprehensively changes the choice of network width and image resolution, so as to make the model performance better, but also save computing resources as much as possible. The experimental results also show that our proposed model obtains the optimal prediction results. Specifically, the loss value of our method is separately 4 less and 2.1 less than that of resnet and efficientnet method.
翻译:目前,应用自动驾驶技术的一个主要挑战是准确预测车辆轨迹,随着计算机技术的蓬勃发展以及电流深度神经网络的出现,预测结果的准确性有所提高,但是,网络和图像分辨率的深度、宽度和广度仍然是限制模型和预测结果准确性的重要原因,本文的主要创新是RESNET网络和高效网络网络的结合,这不仅大大提高了网络深度,而且全面改变了网络宽度和图像分辨率的选择,使模型性能更好,并尽可能节省计算资源,实验结果还表明,我们提议的模型获得了最佳预测结果,具体地说,我们方法的损失价值比Resnet和高效网方法分别少4倍和2.1倍。