Improving Artificial Intelligence Techniques for Classifying the Vessels of the Grand Mosque of Al-Nuri

The Grand Mosque of al-Nuri is a historic mosque that preserves the civilized history of Mosul. It is one of the oldest mosques in the city and contains the Al-Hadba Minaret, which includes the same name as the city of Mosul. This minaret is distinguished by its architectural design and its varied and wonderful decorations. The architectural design of the mosque is unique. The mosque was damaged during the ISIS campaign in Mosul, and the majority of the vessels and relics within it were broken. As a result, it became important to conserve and electronically document the remaining archeological landmarks. So, in this study a hybrid was made between pre-training model to extract features and machine learning methods to classify the dataset of the vessels of the Grand Mosque of al-Nuri. Several pre-processing was applied to the images and then passed to DenesNet201 to extract features and send them to the Extra Tree and Random Forest methods to classify them into pottery and ceramic categories. The results showed that the two hybrid methods outperformed traditional machine learning methods with an accuracy of 98% and 93%, respectively.