---
license: cc-by-nc-4.0
dataset_info:
  features:
  - name: image
    dtype: image
  - name: annotation
    dtype: image
  splits:
  - name: train
    num_bytes: 8683872818.848
    num_examples: 45728
  - name: val
    num_bytes: 1396718238.836
    num_examples: 15358
  - name: test
    num_bytes: 1516829621.65
    num_examples: 4623
  download_size: 12492798567
  dataset_size: 11597420679.334
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: val
    path: data/val-*
  - split: test
    path: data/test-*
---

# Dataset Card for Cloud-Adapter

This dataset card aims to describe the datasets used in the Cloud-Adapter, a collection of high-resolution satellite images and semantic segmentation masks for cloud detection and related tasks.

## Uses

```python
# Step 1: Install the datasets library
# Ensure you have the `datasets` library installed
# You can install it using pip if it's not already installed:
# pip install datasets

from datasets import load_dataset
from PIL import Image

# Step 2: Load the Cloud-Adapter dataset
# Replace "XavierJiezou/Cloud-Adapter" with the dataset repository name on Hugging Face
dataset = load_dataset("XavierJiezou/Cloud-Adapter")

# Step 3: Explore the dataset splits
# The dataset contains three splits: "train", "val", and "test"
print("Available splits:", dataset.keys())

# Step 4: Access individual examples
# Each example contains an image and a corresponding annotation (segmentation mask)
train_data = dataset["train"]

# View the number of samples in the training set
print("Number of training samples:", len(train_data))

# Step 5: Access a single data sample
# Each data sample has two keys: "image" and "annotation"
sample = train_data[0]

# Step 6: Display the image and annotation
# Use PIL to open and display the image and annotation
image = sample["image"]
annotation = sample["annotation"]

# Display the image
print("Displaying the image...")
image.show()

# Display the annotation
print("Displaying the segmentation mask...")
annotation.show()

# Step 7: Use in a machine learning pipeline
# You can integrate this dataset into your ML pipeline by iterating over the splits
for sample in train_data:
    image = sample["image"]
    annotation = sample["annotation"]
    # Process or feed `image` and `annotation` into your ML model here

# Additional Info: Dataset splits
# - dataset["train"]: Training split
# - dataset["val"]: Validation split
# - dataset["test"]: Testing split
```

## Dataset Structure

The dataset contains the following splits:
- `train`: Training images and corresponding segmentation masks.
- `val`: Validation images and corresponding segmentation masks.
- `test`: Testing images and corresponding segmentation masks.

Each data point includes:
- `image`: The input satellite image (PNG or JPG format).
- `annotation`: The segmentation mask (PNG format).

## Dataset Creation

### Curation Rationale

This dataset was created to facilitate the reproduction of Cloud-Adapter.

### Source Data

#### Data Collection and Processing

The dataset combines multiple sub-datasets, each processed to ensure consistency in format and organization:
- Images and annotations were organized into `train`, `val`, and `test` splits.
- Annotations were verified for accuracy and class consistency.

#### Who are the source data producers?

The dataset combines data from various remote sensing sources. Specific producers are as follows:
- WHU (gf12ms, hrc)
- Cloudsen12 dataset
- L8 Biome dataset

## Citation

<!-- If there is a paper or blog post introducing the dataset, the APA and Bibtex information for that should go in this section. -->

**BibTeX:**

[More Information Needed]

**APA:**

Xavier Jiezou. (2024). *Cloud-Adapter: A Semantic Segmentation Dataset for Remote Sensing Cloud Detection*. Retrieved from https://huggingface.co/datasets/XavierJiezou/Cloud-Adapter.

## Glossary [optional]

[More Information Needed]

## More Information

[More Information Needed]

## Dataset Card Authors

This dataset card was authored by Xavier Jiezou.

## Dataset Card Contact

For questions, please contact Xavier Jiezou at xuechaozou (at) foxmail (dot) com.