Multi-scale deep convolutional neural network-based satellite image target detection method

Abstract

The invention discloses a multi-scale deep convolutional neural network-based satellite image target detection method. The method comprises the steps of: collecting a satellite image training data set, and carrying out sample labeling; preprocessing the satellite image training data set; building a multi-scale deep convolutional neural network; inputting the preprocessed training data set into a target detection framework based on the multi-scale deep convolutional neural network to perform training so as to obtain a trained target detection neural network; and inputting a to-be-detected satellite image set, performing target detection by adopting the trained target detection neural network, and outputting an identification result. With the method adopted, the accuracy of the result of thedetection of fine-grained characteristics by the network is improved; the capability of the network in distinguishing different objects is improved; and the effect of the detection of small objects and dense object groups is improved. The method has higher robustness and can effectively improve target detection efficiency and lower hardware requirements.

Description

technical field [0001] The invention relates to the technical field of detecting multi-scale satellite image targets based on convolutional neural networks, in particular to a satellite image target detection method based on multi-scale deep convolutional neural networks. Background technique [0002] In recent years, the target detection algorithm based on deep learning has greatly improved its efficiency, but there are still a series of problems based on satellite image processing. Satellite imagery is a very important resource. It can be used to measure land resources, detect ground conditions and record changes in the ground with a long-term perspective. [0003] However, due to the huge size of satellite images and the relatively small objects in them, and the lack of training data of high-quality satellite images, it is full of challenges to use satellite images for object detection. For example: the detection effect of small objects and dense object groups in satelli...

AI Technical Summary

Problems solved by technology

[0003] However, due to the huge size of satellite images and the relatively small objects in them, the lack of training data for high-quality satellite images, etc., it is challenging to use satellite images for target detection.

For example: the detection effect of small objects and dense object groups in satellite imagery is not satisfactory; existing object detection algorithms lack certain generalization ability for uncommon scales or new image distributions, and because objects may have different orientations and size scales , the limited ratio change of the algorithm will be invalid for the detection of special targets; at the same time, the existing target detection algorithms process the entire image, but for satellite images with hundreds of millions of pixels, it is difficult to have a hardware graphics card memory that can meet such a large demand

Images