Abstract:
Volumetric visual media captures three-dimensional scene geometry and appearance rather than flat images. It includes point clouds and polygonal meshes that sample discrete surfaces, as well as newer formats such as neural radiance fields (NeRF) and 3D Gaussian splats (3DGS) that render continuous, view-dependent scenes. By supporting six-degrees-of-freedom (6 DoF) navigation, these representations enable immersive augmented reality (AR), virtual reality (VR) and mixed reality (MR) experiences in gaming, film, industrial design, remote collaboration and cultural heritage preservation. As volumetric applications proliferate, super-resolution, denoising, completion, artifact removal and temporal interpolation are essential to improve visual fidelity. On the other hand, the sheer volume of 3D content creates storage and delivery bottleneck, prompting industry and academia to develop efficient compression and transmission techniques. Standards bodies such as MPEG WG4 and WG7 are defining interoperable compression frameworks, while limited network bandwidth and resource-constrained devices call for adaptive streaming methods and real-time rendering optimizations. In line with these challenges and ongoing efforts, we invite contributions on volumetric visual media processing, evaluation, transmission, compression, rendering and representation.
Organizers:
Yiling Xu, Jenq-Neng Hwang
Abstract:
Recent advancements in generative AI, foundation models, computational imaging, immersive media, medical imaging, video enhancement techniques such as super-resolution, sustainable streaming, and behavior modeling have substantially broadened the scope and impact of Video Quality Assessment (VQA). In particular, the rise of large language models (LLMs), multimodal learning, and embodied AI agents introduces new paradigms for understanding and assessing video content, user interaction, and context.Traditional quality metrics are increasingly inadequate for these modern use cases, which demand a deeper integration of perceptual, behavioral, semantic, and situational understanding. The evolution of VQA now intersects with high-level cognition, intent-aware evaluation, and content generation, prompting a rethinking of both methodologies and benchmarks.This special session brings together researchers tackling these emerging challenges in multimedia quality and Quality of Experience (QoE) assessment. Topics span diverse application domains such as underwater imaging, medical diagnostics, neural-inspired processing, agent-based interaction, and quality-of-life evaluation. Reflecting ongoing discussions within the Video Quality Experts Group (VQEG), the session aims to highlight novel methods, tools, and datasets that transcend traditional QoE paradigms and address the needs of future multimedia ecosystems.
Organizers:
Wei Zhou, Patrick Le Callet, Cosmin Stejerean
