Mongolian folk painting image recognition framework through CNN-based VGG16 architecture

Mongolian folk painting images have long served as a symbol of the richness of their cultural legacy. These images represent a profound cultural demonstration of the nomadic lifestyle and spiritual beliefs of the Mongolian people. This study presents a predictive framework that uses CNN-based VGG16 model as an image recognition technology to identify Mongolian folk art. The proposed framework consists of multiple parts, such as image source engines for automated data collecting, image classes for Mongolian and Indian traditional paintings, dataset creation, and manual image visualization to guarantee the quality of the datasets. Stability of the dataset is improved by preprocessing methods like pixel value normalization and overfitting minimization. With its deep convolutional neural network architecture, the CNN-based VGG16 model is used for image recognition and shows skill in recognizing intricate visual patterns. The proposed framework is distinguished by a methodical algorithmic flow that includes dataset creation, model training, iterative refinement, and data retrieval. Test accuracy (77.97%) and test loss (0.6828) demonstrate how well the model recognizes Mongolian folk paintings. Extensive results that include feature extraction, ROC and PR curves, and confusion matrix analysis highlight how well the framework achieves a balanced precision-recall trade-off. This study contributes a useful methodology to the growing field of cultural image recognition, highlighting the integration of traditional art with advanced technology.