Ship Detection in Complex Weather Based on CNN
JingXue Chen, Jie Wang, Hua Lu · 2021
Ship detection methods which can be used in all kinds of complex weather is of great significance to the waterborne traffic. This paper uses the algorithm of dark channel estimation to simulate a data set under complex weather conditions naturally by small-scale dataset contain images mostly taken on sunny days. Besides, this paper compares several detectors based on CNN and different structure of feature extraction network. Finally, this paper gets a ship detection model that can achieve great performance under both original small-scale dataset and extended dataset with simulated complex weather condition images.