A vehicle detection using selective multi-stage features in convolutional neural networks
Won‐Jae Lee, Dong-Sung Pae, Dong-Won Kim, Myo Taeg Lim · 2017
Vehicle detection is the most basic and important technology in advanced driver assistant system. Conventional methods do not reflect characteristic information of vehicle images, so they were vulnerable to noise. In order to improve the performance of vehicle detection, this paper proposes a vehicle detection framework using selective multi-stage features in convolutional neural networks. We design the convolutional neural network (CNN) model with 10 layers and use a visualization technique to selectively extract features from the activation feature map in CNN. Our proposed features have the characteristic information of vehicle images and are more robust to noise than traditional appearance based features. We train the Adaboost algorithm using these features to implement a vehicle detector. The result of the experiments proves that our proposed vehicle detection framework has a better performance than other frameworks.