Flickr-Faces-HQ Dataset (FFHQ) dataset: https://github.com/NVlabs/ffhq-dataset The dataset consists of 70,000 high-quality PNG images at 1024×1024 resolution and contains considerable variation in terms of age, ethnicity and image background. It also has good coverage of accessories such as eyeglasses, sunglasses, hats, etc. The images were crawled from Flickr, thus inheriting all the biases of that website, and automatically aligned and cropped using dlib. Only images under permissive… See the full description on the dataset page: https://huggingface.co/datasets/marcosv/ffhq-dataset.

This is a small version of FFHQ with 3143 photos. Flickr-Faces-HQ (FFHQ) is an image dataset containing high-quality images of human faces. It is provided by NVIDIA under the Creative Commons BY-NC-SA 4.0 license. It offers 70,000 PNG images at 1024×1024 resolution that display diverse ages, ethnicities, image backgrounds, and accessories like hats and eyeglasses.

FFHQ 70000张png图片 链接：https://pan.baidu.com/s/1XDfTKWOhtwAAQQJ0KBU4RQ 提取码：bowj

  Flickr-Faces-HQ Dataset (FFHQ)

Flickr-Faces-HQ (FFHQ) is a high-quality image dataset of human faces, originally created as a benchmark for generative adversarial networks (GAN):

A Style-Based Generator Architecture for Generative Adversarial Networks Tero Karras (NVIDIA), Samuli Laine (NVIDIA), Timo Aila (NVIDIA) http://stylegan.xyz/paper

The dataset consists of 70,000 high-quality PNG images at… See the full description on the dataset page: https://huggingface.co/datasets/student/FFHQ.

You can find all information of the dataset in github repository.

https://github.com/NVlabs/ffhq-dataset

Downloading in Google Drive is too slow, so I upload the dataset.

Flickr-Faces-HQ Dataset (FFHQ) - 1024x1024

This is a reupload of FFHQ-1024. Refer to the original dataset repo for more information https://github.com/NVlabs/ffhq-dataset Specifically, this is the images1024x1024 set - faces are aligned and cropped to 1024x1024. Original PNG files were transcoded to WEBP losslessly to save space and packed to WebDataset format for ease of streaming. The original filenames are kept (with different file extension) so that you can match against… See the full description on the dataset page: https://huggingface.co/datasets/gaunernst/ffhq-1024-wds.

Dataset

This dataset was created by Rahul Bhalley

Contents

Extension of the FFHQ dataset (https://github.com/NVlabs/ffhq-dataset) with precise bounding box annotations for eyeglasses detection.

If you find this dataset useful in your research, please consider citing the original dataset and the following papers:

@article{matuzevicius2024diverse,
title={Diverse Dataset for Eyeglasses Detection: Extending the Flickr-Faces-HQ (FFHQ) Dataset},
author={Matuzevi{\v{c}}ius, Dalius},
journal={Sensors},
volume={24},
number={23},
pages={7697},
year={2024},
publisher={MDPI},
url = {https://doi.org/10.3390/s24237697}
}

@article{matuzevicius2024retrospective,
title={A Retrospective Analysis of Automated Image Labeling for Eyewear Detection Using Zero-Shot Object Detectors},
author={Matuzevi{\v{c}}ius, Dalius},
journal={Electronics},
volume={13},
number={23},
pages={4763},
year={2024},
publisher={MDPI},
url = {https://doi.org/10.3390/electronics13234763}
}

The dataset consists of 111,016 high-quality JPG format DeepFake(face swapped) images at a resolution of 512×512, using Flickr-Faces-HQ Dataset (FFHQ). The images are named as "source_target.jpg" e.g. "0_1.jpg" where source image is the "0.png" and target image is the "1.png" from Flickr-Faces-HQ Dataset (FFHQ). This dataset was developed as a part of the research work titled "Uncovering DeepFake Images for Identifying Source-images".

@INPROCEEDINGS{11021847,
 author={Syeda Jannatul Naim and Sarker Tanveer Ahmed Rumee},
 booktitle={2024 27th International Conference on Computer and Information Technology (ICCIT)}, 
 title={Uncovering DeepFake Images for Identifying Source-images}, 
 year={2024},
 pages={2629-2634},
 keywords={Deepfakes;Image analysis;Computer architecture;Security;Information technology;Image reconstruction;Faces;DeepFake;Faceswap;DeepFake image analysis;DeepFake source images},
 doi={10.1109/ICCIT64611.2024.11021847}}

S. J. Naim and S. T. A. Rumee, "Uncovering DeepFake Images for Identifying Source-images," 2024 27th International Conference on Computer and Information Technology (ICCIT), Cox's Bazar, Bangladesh, 2024, pp. 2629-2634, doi: 10.1109/ICCIT64611.2024.11021847.

Dmini/FFHQ-64x64 dataset hosted on Hugging Face and contributed by the HF Datasets community

bitmind/ffhq-jpg dataset hosted on Hugging Face and contributed by the HF Datasets community

heboya8/ffhq-dataset dataset hosted on Hugging Face and contributed by the HF Datasets community

DEPRECATED!!!

Please see this thread for the new versions: https://www.kaggle.com/c/deepfake-detection-challenge/discussion/122786

LSUN-{cat, bedroom, church} [48] and FFHQ [25] datasets

FFHQ Faces Data Set Ported to Kaggle from: https://github.com/NVlabs/ffhq-dataset

Related paper: A Style-Based Generator Architecture for Generative Adversarial Networks Tero Karras (NVIDIA), Samuli Laine (NVIDIA), Timo Aila (NVIDIA) http://stylegan.xyz/paper

This data set only includes 'thumbnail' images resized to (224px by 224px) from original (128px by 128px) size

Large scale image super-resolution is a challenging computer vision task, since vast information is missing in a highly degraded image.

wangx0t/FFHQ dataset hosted on Hugging Face and contributed by the HF Datasets community

The dataset used in the paper is a large dataset of images, including FFHQ, AFHQ-Cat, and LSUN-Church.

The dataset used for the experiments in the paper, including CelebA-HQ, FFHQ, and AFHQ datasets.

Dataset

This dataset was created by Jayanthi Raghavan

Contents

FFHQ 256x256 is frequently used for evaluating unconditional generation ability of generative models (such as GANs and denoising diffusion/score-based models). It's easy to download the original 1024x1024 images from the dataset repo but here's a smaller download because the images are already downscaled.

The .zip file contains 70k images in PNG format and was constructed by downloading the original zip file from the dataset repo, iterating through the images and downsizing them using pillow (bicubic interpolation). If this was useful for you, please drop a message 🤗