Method and device for identifying ship brand character in natural scene based on deep learning

Abstract

The invention relates to a method and a device for identifying a ship brand character in a natural scene based on deep learning. The method comprises the steps of acquiring a video of incoming and outgoing ships at a port; decomposing the video into video streams composed of ship image frames, and classifying the video streams by use of a trained ship classification model to acquire a ship image in a specified type; carrying out ship brand detection on the ship image in the specified type by use of a trained ship brand detection module to acquire a ship brand detection result; and carrying out character identification on the ship brand detection result to obtain the ship brand character in the natural scene.

Description

technical field [0001] The present invention relates to video processing, artificial intelligence, and deep learning, in particular to a method and device for character recognition of ship signs in natural scenes based on deep learning. Background technique [0002] Ship plate recognition is a research problem in the field of target recognition in complex scenes. It is difficult to apply common text recognition methods to the field of ship plate recognition. At the same time, with the rise of deep learning, target recognition in complex scenes has been developed to a large extent, and the recognition accuracy and efficiency are getting higher and higher. However, the field of ship plate recognition has not yet made breakthrough progress, especially Facing the challenge of complex port environment. Deep learning has developed very rapidly in recent years, and its application fields have become more and more extensive. As a method of deep learning, convolutional neural networ...

AI Technical Summary

Problems solved by technology

However, ship text recognition in the natural environment faces many difficulties and challenges: there is no uniform ship plate logo, font specifications are not uniform, ship plate positions are not uniform, complex weather and text occlusion, etc. have brought great challenges to ship plate text recognition. It is a big challenge, so only relying on ship plate text recognition cannot achieve the desired recognition effect

[0004] In recent years, with the globalization of the economy, my country's maritime import and export trade has developed rapidly, and the number of ships entering and leaving the port has also increased sharply. Effective management of ships is prone to illegal berthing, causing ship collision accidents, which will not only cause casualties and property losses, but also cause serious pollution to the marine environment, affect the operational efficiency of the port and cause serious economic losses

Images