randall-lab/celeb-a

Dataset Card for CelebFaces Attributes (CelebA)

Dataset Details

Dataset Description

The CelebFaces Attributes Dataset (CelebA) consists of 202,599 facial images of 10,177 individuals, annotated with 40 binary attributes per image (e.g., smiling, eyeglasses, male/female).

In our repository, we use only the images and attributes, making the dataset suitable for multi-label classification.

Dataset Sources

• Homepage: https://mmlab.ie.cuhk.edu.hk/projects/CelebA.html
• Paper: Liu, Z., Luo, P., Wang, X., & Tang, X. (2015). Deep learning face attributes in the wild. In Proceedings of the IEEE international conference on computer vision (pp. 3730-3738).

Dataset Structure

Total images: 202,599

Attributes: 40 binary labels per image

Splits:

• Train: 162,770 images

• Validation: 19,867 images

• Test: 19,962 images

Image specs: JPEG format, RGB images

Example Usage

Below is a quick example of how to load this dataset via the Hugging Face Datasets library.

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("randall-lab/celeb-a", split="train", trust_remote_code=True)
# dataset = load_dataset("randall-lab/celeb-a", split="test", trust_remote_code=True)
# dataset = load_dataset("randall-lab/celeb-a", split="validation", trust_remote_code=True)

# Access a sample from the dataset
example = dataset[0]
image = example["image"]
attributes = example["attributes"]

image.show()  # Display the image
print(f"Attributes: {attributes}")

Citation

BibTeX:

@inproceedings{liu2015faceattributes, title = {Deep Learning Face Attributes in the Wild}, author = {Liu, Ziwei and Luo, Ping and Wang, Xiaogang and Tang, Xiaoou}, booktitle = {Proceedings of International Conference on Computer Vision (ICCV)}, month = {December}, year = {2015} }