Perceptual Distance

Perceptual distance research aims to quantify how humans and machines perceive the differences between stimuli, such as images or sounds, focusing on developing accurate models that reflect subjective judgments. Current research emphasizes improving the evaluation of these models, particularly using techniques like binomial distribution fitting for two-alternative forced choice data and developing benchmarks to assess the geometric and dimensional understanding of large vision-language models (VLMs). This work is crucial for advancing fields like image compression, virtual reality, and AI, enabling the creation of more realistic and user-friendly technologies.