Image Semantic Segmentation Algorithm Based On Lightweight Attention Mechanism And Improved Encoding And Decoding Structure

Semantic segmentation is the basis of computer vision tasks.Its purpose is to divide the scene into different image regions and assign a corresponding semantic label to each pixel in the scene.This technology is currently widely used in many fields such as Geoscience Information System(GIS),autonomous driving,and medicine.At the same time,the deep learning boom has also promoted the development of semantic segmentation.How to enhance the ability of network feature representation,reduce the loss of image details,and reduce model calculation costs has become the focus of semantic segmentation tasks.To improve the ability of network feature expression and reduce the loss of image details,this paper proposes a bilateral attention network.Among them,the correlation coefficient channel attention module obtains the information between channel responses,and uses this as a weight to weight and fuse the information between each channel,which increases the characterization ability of features.And the position attention module proposed in the decoding structure enhances the spatial information of the feature,and after fusing with high-dimensional features,it has achieved better results.Compared with the existing attention mechanism,the proposed attention module can capture more contextual information,so it can obtain better results.In view of the current problem of the high computational cost of the key-value attention module,a lightweight attention module is proposed,which can further improve network performance while reducing the computational cost.In addition,an improved up-sampling module is proposed for the improper processing of bilinear Interpolation up-sampling in image edge details.