Ship target identification method, computer device, program product and storage medium

Abstract

The invention discloses a ship target recognition method, a computer device, a program product and a storage medium, and provides a ship target recognition normal form method based on transfer learning by adopting a transfer learning method in artificial intelligence and combining an image target recognition technology. The automatic evaluation and integration normal form of the ship target identification transfer learning method in a specific application scene is realized, the diagnosis and optimal selection capability of transfer learning is formed, and the model identification precision is improved. The migration model quality evaluation capability matrix presents association relationships between different migration models and ship sample working condition domains, and a migration method capability evaluation matrix is constructed through three dimensions, namely a migration learning method, ship working conditions and performance indexes, so that a reference basis is provided for model difference comparison; and through a multi-dimensional index system of the capability matrix, the domain migration capability of the base model can be scientifically evaluated and the feature weighting of the base model can be guided. According to the invention, the target identification accuracy is improved.

Description

technical field [0001] The invention relates to ship target recognition technology, in particular to a ship target recognition method, a computer device, a program product, and a storage medium. Background technique [0002] Today, with the increasing degree of globalization, the ocean is an important link for economic and cultural exchanges among countries around the world. As a tool for human beings to develop and utilize the ocean, ships play an irreplaceable role. Ship targets mainly include civilian ships and military ships. Ship target recognition is an important part of modern maritime intelligent monitoring system. Rapid and accurate recognition of various ship targets is of great significance for monitoring maritime traffic, maintaining maritime rights and interests, and improving coastal defense early warning capabilities; in modern warfare, ship Ship target recognition technology is widely used in military fields such as early warning detection, precision guidan...

AI Technical Summary

Problems solved by technology

[0005] (1) The migration model of conventional target recognition only considers the source domain and the target domain from a single dimension, while ignoring the impact of data domains on the model in different dimensions and the sensitivity of the model to specific features

[0006] (2) In a complex real-world environment, there is currently no evaluation system that can make an objective and reasonable evaluation of the transfer learning model, that is, it is impossible to measure the performance of the model from multiple dimensions

[0007] (3) The features learned by the single target recognition model are limited and incomplete, and the accuracy of target recognition is difficult to greatly improve

[0008] (4) A single target recognition model has limited recognition accuracy and is more sensitive to specific features, resulting in a single learned feature and is more sensitive to noise

Images