Image detection method and system based on efficient bidirectional path aggregation attention network

Abstract

The invention provides an image detection method and system based on an efficient bidirectional path convergence attention network. The image detection method based on the efficient bidirectional path convergence attention network comprises the steps of S110, extracting multi-scale features of a to-be-detected sample; S120, selecting preset features in the multi-scale features to carry out feature fusion, and obtaining and outputting a plurality of effective feature prediction maps; S130, performing convolution operation on the plurality of effective feature prediction maps, and predicting the category, the position and the confidence coefficient of the target through a classification regression network; and S140, screening classification results, and outputting a final detection result. According to the image detection method and system provided by the invention, automatic and reliable detection of the image target can be realized. According to the method and system, high-level semantics and spatial information are efficiently fused through IEPAN, so that the capability of capturing multi-scale scattering features of a target by a network is enhanced. The features are refined through a lightweight ERSA module, effective features are distinguished in a self-adaptive mode so as to cope with interference of image complex backgrounds and multiplicative speckle noise, and the false detection rate is reduced.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to an image detection method and system based on an efficient two-way path aggregation attention network. Background technique [0002] Synthetic Aperture Radar (SAR) is an active microwave imaging radar, which can provide continuous and stable earth observation all day and all day, and has been widely used in various fields. With the maturity of SAR technology, there is a large amount of airborne data onboard, which provides sufficient and rich data support for SAR image target detection. [0003] The powerful feature extraction ability and end-to-end structural advantages of deep learning overcome the shortcomings of traditional methods such as cumbersome manual design of features and complex parameter tuning. At present, the target detection algorithm based on convolutional neural network (CNN) has become the mainstream algorithm, especially the successive appearance of...

AI Technical Summary

Problems solved by technology

In the case of small and dense aircraft targets, and weak feature texture and brightness information, YOLOv5s is prone to missed detection problems

Images