Image detection method based on attention mechanism

Abstract

The invention discloses an image detection method based on an attention mechanism, which can determine an interested area in a picture through the attention mechanism, and comprises the following steps: collecting an image, and obtaining an image data set needing to be tested; dividing the images into a verification set and a test set which are independent and not repeated; performing feature extraction on the images in the verification set and the test set to obtain required feature information; adding an SCSE module composed of a channel attention module and a space attention module based on a Darknet53 network model, and obtaining a model of a test image; taking the image features in the verification set as input model parameters; taking the image features in the test set as input model parameters; and inputting the features of the images in the test set to obtain a corresponding test result. According to the experiment, the precision of picture detection can be improved, the detection efficiency can be improved, and the utilization rate of resources is improved.

Description

technical field [0001] The invention is an attention mechanism-based picture feature detection method, which relates to deep learning and target detection technology. Background technique [0002] Since the deep neural network algorithm first shined on the ImageNet dataset, the field of object detection has gradually begun to use deep learning for research. Subsequently, deep models of various structures were proposed, and the accuracy of the data set was repeatedly refreshed. In fact, deep learning models have left traditional methods far behind in classification tasks. The obvious improvement in image classification has also led to the rapid development of the detection field. Target detection is a kind of detection field, which has been widely used in various fields such as traffic monitoring, human-computer interaction, and precision guidance. Target detection methods can be roughly divided into four types, template matching-based methods, knowledge-based methods, met...

AI Technical Summary

Problems solved by technology

[0003] However, when we detect the relevant features of the picture, the previous detection methods are easily affected by the content of the picture, and the accuracy of extracting and detecting the features of important parts of the picture is not high, which affects the analysis and matching of the image.

Images