Shivalinga (Pind) Detection Using Machine Learning Techniques

Shivalinga, a sacred symbol of Lord Shiva, holds immense cultural and religious significance in Hinduism. Recognizing its image using automated image processing techniques presents unique challenges due to variations in material, lighting, occlusions, and similar-looking artifacts. This research proposes a deep learning-based approach to efficiently recognize Shivalinga images from diverse sources, including temple photographs, sculptures, and digital repositories. The study integrates traditional image processing techniques such as edge detection and contour analysis with advanced deep learning models like CNN and YOLO. The primary objective is to develop an accurate and robust recognition system that overcomes texture variations, background noise, and perspective distortions. A comprehensive image preprocessing pipeline is implemented, incorporating grayscale conversion, Gaussian blurring, and adaptive histogram equalization to enhance feature extraction. Feature engineering techniques, including shape-based (Hu Moments, HOG) and texture-based (LBP, Gabor filters) descriptors, are explored to improve classification accuracy. The system's performance is evaluated using standard metrics such as accuracy, precision, recall, F1-score. Experimental results demonstrate an overall accuracy using CNN is 90%. This research serves as a foundational step towards leveraging artificial intelligence for religious artifact recognition, bridging the gap between technology and cultural heritage conservation.