Classification of Boosting Image Performance Via Quality Enhancement and Swin Transformer

With the increasing use of the Internet and social media and the sharing of millions of images daily, the need to classify these images and collect them in groups has increased for easy searching and finding similar and related ones, which in turn facilitates the work of many researches and works in various fields. The study proposes a method to classify images accurately and efficiently by making modifications to the image that improve the accuracy of the image without distorting it or losing its most important features and using the Swin transformer. The Swin transformer effectively captures contextual information at multiple scales, which leads to improved feature extraction. In addition, improving the accuracy of images enhances the quality of the system's work and obtains the expected results. The method we used reached an amazing accuracy rate of over 90%. The new approach outperformed existing methods, indicating its success in recognizing and classifying images. Classification accuracy has important benefits in recognizing images and which category they belong to, thus greatly helping in making more accurate decisions.