UW-Bench
收藏资源简介:
UW-Bench数据集由重庆大学微电子与通信工程学院创建,专注于城市内涝检测,包含7,677张标注图像,来源于监控摄像头和手持移动设备。数据集在多种不利条件下收集,如低光照、强光反射和清水条件,旨在提高模型在实际应用中的泛化能力。创建过程中,通过人工标注确保数据质量,适用于提升城市内涝检测的准确性和效率。
The UW-Bench dataset was developed by the School of Microelectronics and Communication Engineering of Chongqing University, focusing on urban waterlogging detection. It contains 7,677 annotated images sourced from surveillance cameras and handheld mobile devices. The dataset was collected under various adverse conditions such as low-light environments, strong light reflections and clear water scenarios, aiming to improve the generalization ability of models in practical applications. During the creation process, manual annotation was utilized to ensure data quality, making it suitable for enhancing the accuracy and efficiency of urban waterlogging detection.
城市内涝检测数据集 (UW-Bench)
概述
- 数据集名称: Urban Waterlogging Benchmark (UW-Bench)
- 数据集目的: 用于城市内涝检测,特别是在多样化和不利环境条件下。
- 数据集特点: 包含一般样本和困难样本,用于客观评估模型在实际应用中的能力。
数据集详情
- 训练集: 由重庆大学LiVE团队收集和标注。
- 测试集: 由华为提供。
- 数据集链接:
- 训练集: Baidu Drive | Google Drive
引用
@inproceedings{ song2024lsmadapter, title={Urban Waterlogging Detection: A Challenging Benchmark and Large-Small Model Co-Adapter}, author={Suqi Song and Chenxu Zhang and Peng Zhang and Pengkun Li and Fenglong Song and Lei Zhang}, journal = {ECCV}, issue_date = {2024} }

- 1Urban Waterlogging Detection: A Challenging Benchmark and Large-Small Model Co-Adapter重庆大学微电子与通信工程学院 · 2024年



