Deep Learning-Based Object Recognition and Detection for Identifying Champa

Abstract

Object recognition and detection have experienced remarkable advancements with the emergence of deep learning, enabling more precise and efficient identification of complex patterns. This study explores the application of deep learning techniques for recognizing and detecting Champa-related objects using a dataset of 5000 labeled words. Our approach integrates convolutional neural networks (CNNs) and transformer-based architectures to enhance recognition accuracy and localization performance. The dataset consists of diverse Champa-related objects, including inscriptions, sculptures, architectural elements, and cultural artifacts, making it a valuable resource for training deep learning models. By leveraging CNNs for feature extraction and transformer-based models for contextual understanding, we achieve improved object classification and spatial localization. The proposed model is evaluated on multiple benchmark metrics, demonstrating its effectiveness in identifying Champa heritage with high precision and recall. Experimental results indicate that deep learning-based methods significantly outperform traditional computer vision approaches in recognizing and detecting Champa-related objects. The findings highlight the potential of artificial intelligence in supporting cultural heritage preservation and archaeological research. Moreover, our study provides insights into the role of deep learning in digital humanities, offering a framework for future applications in heritage conservation and historical documentation. By utilizing advanced neural network architectures, this research contributes to the growing intersection of AI and cultural studies. The proposed methodology not only facilitates the recognition of Champa artifacts but also paves the way for broader applications in heritage conservation. Future work will focus on refining the model and expanding the dataset to encompass a wider range of Champa-related elements.