Towards versatile multimedia quality assessment for visual communications

With the rapid advancement and increasing demands of multimedia applications in visual communications, the visual quality of multimedia content has emerged as a pivotal factor, profoundly impacting service quality and user experience. Traditionally, visual quality assessment has concentrated on individual modalities such as images, videos, and 3D models, with evaluation models designed separately due to the distinct characteristics of each media type. However, from the perspective of the human vision system, visual quality across these modalities is interconnected and shares a unified perceptual basis. To address this, we introduce a versatile multimedia visual quality assessment framework tailored for visual communications, which unifies quality assessment of images, videos, and 3D models within a single large multi-modal model (LMM). This integrated approach enables simultaneous quality evaluation across all three modalities, effectively harnessing cross-domain knowledge while reducing the inefficiencies and resource overhead of deploying separate models in multimodal communication systems. Experimental results show that our proposed framework, X-QA, delivers robust quality assessment performance across images, videos, and 3D models, establishing a strong technical foundation and opening new possibilities for future visual communication applications requiring sophisticated multimodal quality evaluations.