VEHICLE TYPE DETECTION BASED ON RETINANET WITH ADAPTIVE LEARNING RATE ATTENUATION

Yiliu Xu, Peng He, Hui Wang, Ting Dong, and Pan Shao

Keywords

Learning rate, RetinaNet, focus loss, vehicle type detection, pyramidfeature network

Abstract

Aiming at the problem that the accuracy and speed of the current target detection algorithm cannot be balanced, this paper uses RetinaNet as the basic framework for vehicle type detection and proposes an adaptive learning rate attenuation (ALRA) on the basis of the least squares, which can promote model convergence effectively. The experimental results show that training time cost is reduced and accuracy is improved by using ALRA to train the model.

Important Links:

Go Back