LYNRED Mobility Dataset: Multimodal Detection Subset (LYNRED-MDS)
收藏资源简介:
LYNRED-MDS是由法国格勒诺布尔阿尔卑斯大学与林瑞德公司联合创建的大规模多模态驾驶场景数据集,专注于提升热红外成像在复杂环境下的行人检测性能。该数据集包含4000组严格对齐的RGB-红外图像对,覆盖法国格勒诺布尔地区全年度3.7°C至30°C温度区间,涵盖城市、山区、乡村等多样化驾驶场景,并标注9类道路使用者及遮挡状态。数据通过车顶双立体相机系统采集,采用触发同步机制确保模态对齐,特别关注极端天气与低能见度边缘案例。本数据集旨在推动高级驾驶辅助系统在热成像模态上的泛化能力研究,为解决夜间、雾天等恶劣条件下的行人检测难题提供关键基准。
LYNRED-MDS is a large-scale multimodal driving scene dataset jointly developed by Grenoble Alpes University (France) and Lynred Company, focused on enhancing the pedestrian detection performance of thermal infrared imaging in complex environments. It contains 4,000 strictly aligned RGB-thermal image pairs, spanning the full-year temperature range of 3.7°C to 30°C in the Grenoble region of France, and covers diverse driving scenarios such as urban, mountainous, and rural areas. The dataset is annotated with 9 categories of road users and their corresponding occlusion states. Data collection is conducted via a roof-mounted dual stereo camera system, utilizing a trigger synchronization mechanism to ensure strict modal alignment, with particular emphasis on extreme weather and low-visibility edge cases. This dataset is designed to advance research on the generalization capability of Advanced Driver Assistance Systems (ADAS) leveraging thermal imaging modalities, serving as a critical benchmark for addressing the challenges of pedestrian detection under harsh conditions including nighttime and foggy environments.

- 1Descriptor: LYNRED Mobility Dataset Multimodal Detection Subset (LYNRED-MDS)法国国家信息与自动化研究所·格勒诺布尔阿尔卑斯大学中心; 林瑞德公司; 格勒诺布尔阿尔卑斯大学·CNRS·GIPSA实验室 · 2026年




