A Modern Approach to Monument Identification using Deep Learning Techniques

Sophisticated management techniques and exact monitoring are now required for the preservation, study, and comprehension of historical places, particularly for the identification of monuments. This study introduces a fresh application of artificial intelligence (AI) in monument identification, employing the advanced object detection model YOLOv8. By utilizing image and video data, our approach aims to revolutionize the study, preservation, and comprehension of historical sites. We trained our system by assembling a comprehensive dataset of heritage temples and meticulously labelling each. Leveraging Neural Networks, our system gains intelligence, with Convolutional Neural Networks (CNNs) and YOLOv8 employed to identify individual monuments across diverse historical contexts. To further ensure the efficacy of the system, we employ advanced metric assessment methodologies to regularly monitor critical performance metrics. As a result, WandB (Weights and Biases) and other external platforms are no longer required, providing real-time insights into detection accuracy, model behaviour, and potential biases. The agile approach facilitates rapid adjustments and enhancements, resulting in an efficient and reliable monument detection system. Additionally, our model incorporates a Text-to-Speech (TTS) system, converting translating the names of recognized heritage sites, their historical background, and other pertinent information into natural language.