An Efficient and Controllable Capsule Network for Classification of Maritime Moving Targets

The accurate classification of maritime moving targets under complex sea conditions is a very important task. And it plays an important role in ensuring the safety of navigation and safeguarding maritime rights and interests and security, and has a wide range of applications in military and civilian fields. At present, there are many methods for related research, but there are still problems such as a large number of parameters and a high computational cost. Based on this, this article proposes a new efficient and controllable capsule network method for maritime moving target classification research. First, a non-iterative attention routing is constructed to reduce the number of parameters and weaken the dependence of the previous capsule layer on the next capsule layer in the routing process. Second, a controllable routing algorithm is proposed, which can reduce the number of routing and computational cost dynamically according to the complexity of samples and has strong flexibility. Finally, the effectiveness of the proposed method in the classification of maritime moving targets is fully demonstrated by experiments on different datasets.