Abstract
Indian folk paintings have a rich mosaic of symbols, colors, textures, andstories making them an invaluable repository of cultural legacy. The paperpresents a novel approach to classifying these paintings into distinct artforms and tagging them with their unique salient features. A custom datasetnamed FolkTalent, comprising 2279 digital images of paintings across 12different forms, has been prepared using websites that are direct outlets ofIndian folk paintings. Tags covering a wide range of attributes like color,theme, artistic style, and patterns are generated using GPT4, and verified byan expert for each painting. Classification is performed employing theRandomForest ensemble technique on fine-tuned Convolutional Neural Network(CNN) models to classify Indian folk paintings, achieving an accuracy of91.83%. Tagging is accomplished via the prominent fine-tuned CNN-basedbackbones with a custom classifier attached to its top to perform multi-labelimage classification. The generated tags offer a deeper insight into thepainting, enabling an enhanced search experience based on theme and visualattributes. The proposed hybrid model sets a new benchmark in folk paintingclassification and tagging, significantly contributing to cataloging India'sfolk-art heritage.