Table 1 A brief outline of state-of-the-art methods for mythological and cultural image classification.
References | Key features |
|---|---|
Sherly et al.31 | • Utilized object detection techniques like YOLOv5 and ensemble models • Dataset consisted of Hindu and Christian divine images |
Sasithradevi et al.32 | • Proposed a custom KolamNetv2 architecture comprising of EfficientNet and various attention modules • It aimed to classify various artistic styles of rangoli commonly found in Tamil culture |
Sasithradevi et al.33 | • Proposed MonuNet architecture for classification of cultural heritage in Kolkata • Utilized dense channel attention modules and PSCA modules to extract key architectural details from the images |
Babić et al.34 | • Compares four machine learning algorithms using features from eleven pre-trained deep learning models on a small cultural heritage image dataset • Used DenseNet121, EfficientNetB0 and NASNetMobile architectures for feature extraction |
Gao et al.35 | • Proposed Convolutional Neural Network Attention Retrieval Framework (CNNAR Framework) to classify various Chinese diaspora architectural styles • Tested the proposed architecture on Paris500K and Corel5K datasets to achieve considerable results |
