TSOD10K
收藏资源简介:
TSOD10K是首个大规模的交通显著目标检测数据集,由湖南师范大学信息科学与工程学院、南开大学人工智能学院和湖南大学机器人学院共同创建。该数据集包含13753张车辆捕获的图像,并带有像素级的精确注释。TSOD10K覆盖了各种真实世界交通场景,包括城市交叉口、高速公路、乡村道路和停车场,涵盖了雨、雪、雾、晴、低光等不同的天气/光照组合。数据集旨在支持动态风险分层下游服务,特别编码了视觉显著性和隐含的风险语义。
TSOD10K is the first large-scale traffic salient object detection dataset, co-developed by the College of Information Science and Engineering of Hunan Normal University, College of Artificial Intelligence of Nankai University, and College of Robotics of Hunan University. The dataset contains 13,753 vehicle-captured images with pixel-level precise annotations. TSOD10K covers diverse real-world traffic scenarios, including urban intersections, highways, rural roads, and parking lots, and encompasses various weather and lighting conditions such as rain, snow, fog, sunny weather, and low-light environments. The dataset is designed to support downstream dynamic risk stratification services and specially encodes visual saliency and implicit risk semantics.

- 1Salient Object Detection in Traffic Scene through the TSOD10K Dataset湖南师范大学信息科学与工程学院, 南开大学人工智能学院, 湖南大学机器人学院 · 2025年



