11篇论文被ACM MM 2026录用

TL;DR - A roundup of 11 ACM MM 2026 accepted papers spanning multimodal perception, generative imaging, autonomous agents, remote sensing, medical imaging, and visual benchmarks. The collection highlights new datasets, evaluation frameworks, and efficient or training-free methods.

• New benchmarks target satellite multi-object tracking, RGB-D video saliency, mobile GUI-agent memory, and classical-painting reconstruction.
• Agent-focused work includes multi-agent audio-visual segmentation and an evaluation of memory failures across 11 mobile GUI agents.
• New methods address open-vocabulary detection, federated long-tailed learning, diffusion-based lighting adaptation, and hyperspectral change detection.
• Other contributions cover unified CT reconstruction and scalable multi-view, multi-label feature selection.