NWPU-Traffic
收藏资源简介:
NWPU-Traffic是由西北工业大学构建的大规模遥感交通目标分割数据集,旨在解决现有数据集在目标类别、场景多样性和图像时效性方面的局限。该数据集包含1,479张高分辨率遥感图像,涵盖汽车、飞机、船舶和火车四类交通目标,实例总数达31,628个,数据来源于全球7个国家49个城市的多样化环境。数据集通过专业标注团队进行实例级精细标注,并严格进行质量检验,确保了标注的精确性与一致性。该数据集主要应用于遥感图像语义分割领域,为智能交通、城市规划及应急响应等任务提供关键数据支撑,以提升模型在复杂真实场景下的泛化与分割能力。
NWPU-Traffic is a large-scale remote sensing traffic object segmentation dataset constructed by Northwestern Polytechnical University, which aims to address the limitations of existing datasets in terms of object categories, scene diversity and image timeliness. This dataset contains 1,479 high-resolution remote sensing images, covering four types of traffic objects including cars, airplanes, ships and trains, with a total of 31,628 instances. The data is collected from diverse environments across 49 cities in 7 countries worldwide. The dataset undergoes fine-grained instance-level annotation by professional annotation teams and strict quality inspection, ensuring the accuracy and consistency of annotations. This dataset is mainly applied in the field of remote sensing image semantic segmentation, providing critical data support for tasks such as intelligent transportation, urban planning and emergency response, so as to improve the generalization and segmentation capabilities of models in complex real-world scenarios.
NWPU-Traffic 数据集
地址: https://github.com/CVer-Yang/NWPU-Traffic
简介: NWPU-Traffic 是首个综合遥感数据集,覆盖了汽车、船只、飞机和火车四类交通目标。
数据集获取: 该数据集已公开发布,可通过以下链接下载: https://drive.google.com/drive/folders/1LobucN-znqknQrHJs0_jZcqJfAaqt1qz?hl=zh-cn
相关代码: 该数据集对应的算法代码已发布,代码基于 GeoSeg(https://github.com/WangLibo1995/GeoSeg)实现。

- 1A Large-Scale Dataset and a New Method for RemoteSensing Traffic Object Segmentation西北工业大学·人工智能学院; 西北工业大学·计算机学院; 西北工业大学·软件学院 · 2026年



