Semantic segmentation of night vision images for unmanned vehicles based on deep learning

In order to assist unmanned vehicles in understanding scene of night vision images, detecting and identifying surrounding environment more quickly and accurately at night, a semantic segmentation method of unmmanned vehicle night vision images based on convolution-deconvolution neural network is proposed, which uses deep learning to segment scenery semant of night vision images. Convolution-deconvolution neural network is constructed by adding deconvolution network to traditional convolutional neural network, without selecting feature manually. By learning and training pixels-to-pixels, image semantic segmentation model can be obtained. The model can be used to predict scene semantic category of each pixel in night vision image, realizing environment perception of unmanned vehicles at night, which is import for automatic driving at night. Experimental results show that this method has good accuracy and real-time performance, and average IU reaches 68.47.