| license: cc-by-nc-4.0 | |
| language: | |
| - zh | |
| [github](https://github.com/yuyijiong/remote_sense_image_quality_inspection) | |
| [paper](https://arxiv.org/abs/2307.11965) | |
| 使用swin v2模型检测遥感图像中是否包含以下9种类型的质量问题： | |
| "0": "云", | |
| "1": "阴影", | |
| "2": "拉花", | |
| "3": "模糊", | |
| "4": "光谱溢出", | |
| "5": "扭曲", | |
| "6": "拼接痕迹", | |
| "7": "拼接错误", | |
| "8": "条状噪声" | |
| 模型输出为9维向量，每一维的值代表图片中存在此类质量问题的概率，概率大于50%视为存在此类质量问题。 | |