Deep learning-based ship board character accurate detection method

Abstract

The invention discloses a deep learning-based ship board character accurate detection method, and the method comprises the steps of obtaining a target image in real time, inputting the target image into a ship detection classification model, and obtaining a ship image containing ship board characters, wherein the ship detection and classification model detects and classifies an input target imageto obtain a first image containing ship board characters, positions a ship in the first image to obtain a positioning position of the ship in the first image, and intercepts the first image accordingto the positioning position to obtain a ship image containing the ship board characters; preprocessing the ship image to obtain a preprocessed image; inputting the preprocessed image into a ship boardcharacter recognition model to obtain ship board character information, wherein the shipboard character recognition model is used for performing character text recognition on an input preprocessed image and outputting shipboard character information. According to the technical scheme of the invention, the method achieves the precise recognition of the small ship side characters, and improves therecognition accuracy and detection efficiency of the ship side characters.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a method for accurately detecting shipboard characters based on deep learning. Background technique [0002] In the past, the detection of the hull number was basically judged by the human eye. When there are many ships or the distance between the ships is long, the human eye can only capture the hull number of one ship at a time, and when the human eye is in a long-distance observation, The ship's serial number often appears unclear and difficult to distinguish; in this case, the method of using this human eye to detect the ship's serial number is often poor in accuracy and low in efficiency. [0003] With the advent of the era of big data and the continuous optimization of computer hardware performance, methods based on deep learning are constantly developing. In the field of computer vision, convolutional neural network (CNN), as a deep learning method, has shown str...

AI Technical Summary

Problems solved by technology

However, in the prior art, the initially collected image frames are directly used as input parameters for training, but in the actual scene, due to the inability to ensure the appropriate range of acquisition distance when acquiring pictures, the target ship may be smaller in the picture, and at the same time The characters on the side of the ship are small, which ultimately makes the detection effect of the trained ship recognition model unsatisfactory

Images