Organizers 组织者

Abstract / 摘要

• With the rapid advancement of artificial intelligence and computer vision technologies, open-world object detection and image segmentation for real-world scenarios have become important research topics in both academia and industry. Compared with closed-set settings, real-world visual perception tasks face more challenging conditions, including open categories, distribution shifts, limited annotated samples, as well as occlusion, motion blur, low-light environments, and adverse weather. Conventional detection and segmentation methods still suffer from limited generalization, robustness, and continual adaptation capabilities under such complex conditions.
• This special session focuses on the key theories, advanced methodologies, and representative applications of open-world detection and segmentation in real-world scenarios. It will provide a platform for discussing frontier topics such as open-set recognition, cross-modal perception, zero-shot and few-shot learning, continual learning, and domain adaptation. The session aims to bridge academic research and industrial applications, fostering the development and practical deployment of intelligent visual perception technologies for open-world environments.

随着人工智能与计算机视觉技术的快速发展，面向真实场景的开放世界目标检测与图像分割逐渐成为学术界和产业界关注的重要方向。相比封闭场景设定，真实环境中的视觉感知任务通常面临类别开放、分布变化、样本稀缺以及遮挡、模糊、低光照和极端天气等复杂挑战，传统检测与分割方法在泛化能力、鲁棒性和持续适应能力方面仍存在明显不足。本专题聚焦开放式检测与分割在真实场景中的关键理论、核心方法与典型应用，围绕开放类别识别、跨模态感知、零样本与少样本学习、增量学习、域适应等前沿问题展开交流，旨在搭建一个面向学术研究与产业应用融合的研讨平台，推动开放场景视觉智能技术的发展与落地。

Topics 主题征稿范围

The following topics are within the scope of this special session, but are not limited to: 以下为本分论坛征稿范围，但不限于：

• Open-World Object Detection and Image Segmentation
  开放场景下的目标检测与分割
• Cross-Modal Visual Understanding for Open-World Scenarios
  面向开放场景的跨模态视觉理解
• Domain Adaptation, Zero-Shot and Few-Shot Learning for Vision Tasks
  面向视觉任务的域适应、零样本/少样本学习
• Visual Perception under Occlusion, Blur, Low-Light, and Adverse Weather
  遮挡、模糊、低光照、极端天气等复杂条件感知
• Learning under Low-Resource and Cost-Effective Annotation Settings
  低资源、低成本标注下的模型学习
• Continual Learning, Incremental Learning, and Adaptive Model Updating
  增量学习、持续学习与模型自适应更新
• Large Foundation Model Driven Open-World Detection and Segmentation
  大模型驱动的开放世界检测与分割方法
• Applications of Open-World Detection and Segmentation in Real-World Scenarios
  真实场景应用下的开放检测与分割（自动驾驶、工业瑕疵和医学影像等）
• Evaluation Metrics and Benchmarks for Open-World Detection and Segmentation
  开放世界检测与分割评测指标与基准
• Large-Scale Open-World Datasets for Real-World Scenarios
  真实场景大规模开放数据集构建