Design-Led Research of Big Data and Artificial Intelligence Practices in Digital Media

In this research, we delve into the utilization of big data and artificial intelligence (AI) techniques within the realm of visual communication (VISCOM) art design. The focal points are on layout identification, element grouping, and audience assessment. In a regulated setting, by leveraging convolutional neural networks (CNNs) and the STING algorithm, we handled standardized design elements like logos, posters, scenes, text, and packaging. The clustering approach identified an average of 735 design elements with an accuracy exceeding 82%• For scene recognition, CNN algorithms demonstrated accuracy between 73 % and 89%, with facial recognition achieving over 94% accuracy, demonstrating the effectiveness of AI in complex tasks. Key performance metrics, including precision, recall, and F1 scores, averaged 85%, 82%, and 83%, respectively, reflecting balanced performance in accuracy and sensitivity. Evaluations by a panel of experts confirmed the practical applicability and creative potential of these methods. This research highlights the significant role of AI and big data in enhancing the efficiency and precision of VISCOM design, offering innovative solutions to traditional challenges in visual communication.