Ship target detection method under resource limited condition

Abstract

The invention discloses a ship target detection method under a resource limited condition, and the method comprises the steps: inputting a satellite image into a first region screening network, and obtaining a ship target candidate region image; inputting the ship target candidate region image into a pre-trained ship target detection model for target detection to obtain a preliminary target detection result; performing duplicate removal on the preliminary target detection result by using an NMS method to obtain a target detection result, wherein the ship target detection model under the resource limited condition comprises a second region screening network, a network connection layer and a target detection network which are cascaded in sequence. The method can achieve the quick detection of the ship target under the condition of no GPU, and solves a problem of quick detection of the target under the condition of limited resources.

Description

technical field [0001] The invention belongs to the technical field of image processing, and more specifically relates to a ship target detection method under the condition of limited resources. Background technique [0002] In recent years, with the advancement of aerospace technology, the means of remote sensing image acquisition have become increasingly mature, and the resolution of images, including time resolution, spatial resolution, radiation resolution and spectral resolution, is constantly improving. At present, remote sensing has broken through the bottleneck of data acquisition and is moving towards a new stage of comprehensive application, laying a data foundation for the extraction of marine and nearshore targets. As an important marine target, ships are the key targets of maritime monitoring and wartime strikes. Real-time and effective acquisition of basic information of ships is of great significance in both civil and military fields. In the civilian field, i...

AI Technical Summary

Problems solved by technology

[0005] The shortcomings of existing ship detection methods mainly include: the computational bottleneck of CNN mainly lies in the high-dimensional fully connected layer, which not only has many parameters, but also has high computational complexity, which is easy to lead to overfitting, and requires the same size of image input, and the commonly used target detection algorithm Using a general convolution kernel, the detection effect of the target with a large aspect ratio is poor, resulting in poor detection accuracy

The most important current object detection method based on deep learning requires GPU acceleration, and there are relatively few researches and applications under resource-constrained conditions.

Images