DESCULPT: Indian Temple Sculpture Iconography

Abstract

India is a land comprising of a lot a rich culture and heritage. There are a lot of rare and architectural sculptures which have been discovered and found in the recent years in India. Entity recognition of Indian sculptures can be viewed as one of the very challenging and difficult problems occurring in the very field of image recognition, classification as well as identification. In order to get this project started, it is constructed on a manually formed database con- sisting of pictures of Lord Ganesha and Lord Shiva Linga. In this work, Lord Ganesha and Lord Shiva Linga as the two entities which have to classified. Every image consists of a key characteristic which differentiates it from the other. Ori- entations, angles, size, colours of the images all play an important role in the processing of the model. The images have been taken from various and different angles, sizes, colours and orientations. The model will be trained on the image dataset consisting of the entities in Indian sculpture, in return signifying the rich heritage and culture. Deep learning plays a very important role in image classifi- cation and recognition. The effectiveness and accuracy of the suggested model can be evaluated using Support Vector Machine and Convolutional Neural Net- work respectively. Also many other important libraries such as NumPy, mat- plotlib , OpenCV have also been used in this process of image recognition and classification. The model which this project has helped trained has helped us reach an accuracy of 91% with the help of Deep Learning and Convolutional Neural Network along with the Sequential Model.