Visual Interestingness Evaluation using AI/ML

The interestingness of visual contents such as images/videos are determined by a multitude of factures such as aesthethic appeal, scene composition, color palettes, novelty of the subject etc. The evaluation of image interestingness represents a crucial yet challenging aspect of visual content analysis, particularly in the era of vast digital image collections and social media. This research seeks to explore how Artificial Intelligence (AI) and Machine Learning (ML) techniques can be harnessed to develop robust methods for assessing the subjective concept of image interestingness. By leveraging deep learning models, we aim to automatically identify and rank images based on their ability to capture viewer attention and engagement. This research will investigate the integration of contextual cues, aesthetic principles, and user interaction data to enhance the accuracy and reliability of interestingness assessments, addressing both technical challenges and perceptual nuances in evaluating visual content.

Through empirical validation and comparative analysis against human perception studies and existing metrics, this research aims to advance the state-of-the-art in image interestingness evaluation using AI/ML. By establishing a framework that combines computational efficiency with human-like perceptual judgment, we anticipate contributing to applications ranging from content curation and recommendation systems to visual storytelling and digital marketing strategies.