Flood Area Segmentation

The dataset contains images of flood hit areas and corresponding mask images showing the water region.

There are 290 images and self annoted masks. The mask images were created using [Label Studio](https://labelstud.io/), an open source data labelling software. The task is to create a segmentation model, which can accurately segment out the water region in a given picture of a flood hit area.

Such models cane be used for flood surveys, better decision-making and planning. Because of less data, pre-trained models and data augmentation may be used.

#### Naviagate Dataset:
1. **Image**: Folder containing all the flood images.
2. **Mask**: Folder containing all the mask images.
3. **metadata.csv**: A csv file mapping the image name with its mask.

本数据集包含洪水受灾区域的图像及其对应的显示水域区域的掩膜图像。数据集共包含290张图像及自标注的掩膜。掩膜图像系采用[Label Studio](https://labelstud.io/)（一款开源的数据标注软件）创建。任务目标是构建一个分割模型，该模型能够准确地将洪水受灾区域图像中的水域区域进行分割。此类模型可用于洪水调查、决策优化与规划。鉴于数据量有限，可考虑使用预训练模型和数据增强技术。#### 数据集导航：
1. **图像**：包含所有洪水图像的文件夹。
2. **掩膜**：包含所有掩膜图像的文件夹。
3. **metadata.csv**：一个CSV文件，用于将图像名称与其对应的掩膜进行映射。