Multimodal Quality

Multimodal quality assessment focuses on automatically evaluating the perceptual quality of data encompassing multiple sensory modalities, such as audio, visual, and text information, often in the context of low-quality or distorted inputs. Current research emphasizes developing robust and generalizable models, including large multimodal models (LMMs) and quality-aware fusion frameworks, to accurately predict human perception of quality across diverse datasets and applications. This field is crucial for improving the quality of various technologies, from image and video processing to biometric recognition and 360° media, by providing objective measures that align with human subjective experience. The development of standardized benchmarks and datasets is also a key area of ongoing work.