Bricker
收藏资源简介:
Bricker是由上海交通大学与华为消费者业务集团联合构建的一个面向闪烁条带修复任务的合成与真实配对的多帧RAW数据集。该数据集通过基于光线追踪的物理仿真系统和自动化多曝光捕获工具构建,涵盖了多种显示类型、驱动策略和拍摄条件,旨在提供高质量的曝光括号数据以支持多帧修复研究。数据集包含合成与真实两个子集,其中合成数据模拟了复杂的闪烁条带形态及其随曝光设置的变化,真实数据则捕获了实际拍摄中难以建模的非理想因素,有效弥补了仿真与真实分布之间的域差距。该数据集主要应用于计算机视觉和图像处理领域,特别针对屏幕捕获图像中的闪烁条带去除问题,为多帧RAW修复模型的训练与评估提供了关键数据基础。
Bricker is a synthetic and real paired multi-frame RAW dataset for flickering stripe restoration tasks, jointly constructed by Shanghai Jiao Tong University and Huawei Consumer Business Group. This dataset is built via a ray tracing-based physical simulation system and automated multi-exposure capture tools, covering diverse display types, display driving strategies and shooting conditions, aiming to provide high-quality exposure bracketing data to support multi-frame restoration research. The dataset consists of two subsets: synthetic and real. The synthetic data simulates complex flickering stripe patterns and their variations with exposure settings, while the real data captures non-ideal factors that are difficult to model in actual shooting, effectively bridging the domain gap between simulation and real-world distributions. This dataset is mainly applied in the fields of computer vision and image processing, specifically targeting the flickering stripe removal task in screen-captured images, providing a critical data foundation for the training and evaluation of multi-frame RAW restoration models.





