A remote sensing image target detection method and system based on lightweight distillation network

Abstract

The invention relates to a remote sensing image target detection method and system based on a lightweight distillation network. The method includes: acquiring a remote sensing image to be detected; inputting the remote sensing image to be detected into a pre-established detection network, and obtaining the pre-established The initial prediction result of the remote sensing image to be detected outputted by the detection network; the initial prediction result of the remote sensing image to be detected is up-sampled to the image size of the remote sensing image to be detected, and the final prediction result of the remote sensing image to be detected is obtained; The technical solution provided by the invention is oriented to high-resolution remote sensing images, and realizes the advantages of small amount of model parameters, good generalization ability, high detection recall rate, accurate detection of multi-scale objects, and the like.

Description

technical field [0001] The invention relates to the technical field of remote sensing image processing, in particular to a remote sensing image target detection method and system based on a lightweight distillation network. Background technique [0002] In recent years, the leapfrog development of deep learning has promoted great progress in the field of computer vision, making the performance of many computer vision tasks reach an unprecedented height. Although the deep network model has achieved good performance in many tasks, it is only an experiment and attempt at the scientific research level. If it is transplanted to embedded or mobile devices in consideration of practical applications, it will be greatly affected. Aspect constraints: 1) The amount of model parameters is huge; 2) The model has a large amount of calculation; 3) The power consumption is large. [0003] For embedded or mobile devices with limited hardware resources, the direct application of deep neural ...

AI Technical Summary

Problems solved by technology

Although the deep network model has achieved good performance in many tasks, it is only an experiment and attempt at the scientific level. If it is transplanted to embedded or mobile devices in consideration of practical applications, it will be greatly affected. Aspect constraints: 1) The amount of model parameters is huge; 2) The model has a large amount of calculation; 3) The power consumption is large

Images