Visibility Analysis Based on Deep Learning
Zhen Mao, Xiaoyan Chen, Xiaoning Yan, Shuangwu Zheng · Proceedings of International Conference on Artificial Life and Robotics · 2022
In recent years, visibility analysis through deep learning processing and analysis of video images for different places has become a hot research topic that attracts people's attention.A new deep learning model (A-VGGNet) is proposed to evaluate the visibility of real scenes.The model is constructed on the basis of the VGG classification model, and the classification accuracy of the deep learning model is improved by adding an attention mechanism.The experimental results show that the training success rate is 97.62%, the verification set test accuracy rate is 75.05%, and the test set classification accuracy rate is 85.05%.The proposed model has a good effect on the accuracy evaluation and classification of visibility.