WeatherProofExtra
收藏资源简介:
WeatherProofExtra数据集由富士通研发中心有限公司创建,包含160个场景,涵盖雨、雪和雾等恶劣天气条件。该数据集通过添加恶劣天气和超分辨率数据增强方法,模拟了各种不利天气条件,旨在提高模型在恶劣天气下的语义分割性能。创建过程中,数据集通过对比测试数据与训练数据的分辨率差异,采用超分辨率方法缩小差距。该数据集主要应用于解决恶劣天气条件下的语义分割问题,支持模型在复杂环境中的准确识别和分割。
The WeatherProofExtra dataset was developed by Fujitsu Research and Development Center Co., Ltd. It consists of 160 scenarios covering adverse weather conditions including rain, snow, and fog. This dataset leverages adverse weather simulation and super-resolution data augmentation methods to simulate a wide range of harsh weather conditions, with the objective of enhancing the semantic segmentation performance of models under adverse weather environments. During its development, the dataset addresses the resolution mismatch between test and training data by applying super-resolution techniques to narrow this gap. Primarily designed to tackle semantic segmentation challenges under adverse weather conditions, this dataset enables models to perform accurate object recognition and segmentation in complex real-world environments.




