| The research presented focuses on developing an efficient software system for retrieving and categorizing fine art images in museums and art galleries.With the digitization of art collections,there has been a growing demand for methods that can quickly analyze and organize these collections based on their artistic styles.The proposed technique consists of two phases aimed at improving the precision of style classification.In the first phase,the input image is divided into five sub patches.Each patch is then individually classified using a deep convolutional neural network(CNN)that has been trained specifically for this purpose.The second phase involves a decision-making module that utilizes a shallow neural network.This network is trained using probability vectors obtained from the first-phase classifier.The results from each of the five patches are combined in this phase to deduce the final style classification for the input image.One key advantage of this approach is that the second phase is trained independently of the first phase,using probability vectors instead of images.This helps compensate for any potential errors made during the first phase,leading to improved accuracy in the final classification.To evaluate the proposed method,six different pre-trained CNN models,namely AlexNet,VGG-16,VGG-19,GoogLeNet,ResNet-50,and InceptionV3,were employed as the first-phase classifiers.The second-phase classifier was implemented as a shallow neural network.Three standard art datasets were used to perform experimental trials,and the results demonstrated that the suggested technique significantly outperformed existing methods in terms of accuracy and precision in style classification.Overall,the research contributes to the development of effective software systems for the analysis and categorization of fine art images,making them more accessible to the general public through digital platforms. |
