Ship Target Identification via Bayesian-Transformer Neural Network

Ship target identification is of great significance in both military and civilian fields. Many methods have been proposed to identify the targets using tracks information. However, most existing studies can only two or three types targets, accuracy needs be further improved. Meanwhile, they do not provide a reliable probability result under high-noise environment. To address these issues, Bayesian-Transformer Neural Network (BTNN) complete ship task The aim research improving ability enhance maritime situation awareness strengthen protection traffic safety. Firstly, Encoder (BTE) module that contains four different Encoders used extract discriminate features tracks. Then, Bayesian fully connected layer SoftMax classification. Benefiting from superiority neural network, BTNN result, which captures aleatoric uncertainty epistemic uncertainty. experiments show method successfully nine targets. Compared with traditional methods, increases by 3.8% 90.16%. In addition, compared non-Bayesian Transformer Network, more