Perceptual Similarity

Perceptual similarity research aims to quantify how humans judge the likeness of visual stimuli, bridging the gap between subjective experience and objective measurement. Current efforts focus on developing and refining metrics that accurately reflect human perception, employing deep learning models—including convolutional neural networks and transformers—to extract and compare image features, often leveraging self-supervised or contrastive learning approaches. These advancements have implications for various fields, improving image quality assessment, enhancing content retrieval systems, and providing more nuanced analyses of algorithmic bias in computer vision.