Deep Learning Multiclassification Model: Recognizing Monuments
Yukta Nagpal, Varun Jindal, Vinay Kukreja, Satvik Vats, Rishabh Sharma
Abstract
Dubai is one of the places that tourists visit due to its monuments and different visiting points. In the past, many deep learning models have been developed to classify the monuments in Dubai. This is one of the research areas in today’s environment. The current study targets to classify different Dubai monuments. A total of 9500 images have been collected and preprocessed. The dataset has been divided into 8 different cultural monuments. The 8 monuments that are used for collecting images are Grand Mosque, Burj Khalifa, Burj Nahar, Burj AI Arab Jumeriah, Bastakiya Quarters, Infinity Tower, Jumeirah Mosque, and Dubai Museum. CNN model was used to classify the Dubai monuments. The best accuracy achieved for the multiclassification of heritage monuments is 97.97% with the macro average recorded being 89.59%. The best F1 score and precision in multiclassification is 93.05%. In the future, hybrid multiclassification models can be used for better accuracy and F1 score. More primary images can be collected so that better results can be achieved.