The Depth of Vehicle Recognition Based on Neural Network
Studies the recognition of vehicle types based on deep learning methods. Deep neural network is trained to classify automobile images which are shot from different angles with complex background. Uses the cutting-edge Deep Learning architecture, Caffe and a powerful computational platform, GPU. VGG16 network and AlexNet network are trained and tested for this task respectively. Moreover, applies the classical algorithm, K-Nearest Neighbor for comparison. The result suggests that VGG16 network outperform other methods by a big mar-gin with the accuracy of 97.58%.
Deep LearningVehicle RecognitionConvolutional Neural Network
NETL
NSTL
万方数据
维普
