/ffhq-features-dataset

Gender, Age, and Emotion for Flickr-Faces-HQ Dataset (FFHQ)

Primary LanguageShellOtherNOASSERTION

Gender, Age, and Emotions extracted for Flickr-Faces-HQ Dataset (FFHQ)

License CC Format JSON Images 70000

Teaser image

This dataset provides various information for each face in the Flickr-Faces-HQ (FFHQ) image dataset of human faces.

The dataset consists of 70,000 json files, each corresponding to a face. For each face, it contains some of the following information:

  • head pose
  • gender
  • age
  • moustache, beard, and sideburns probability
  • glasses
  • emotion: anger/contempt/disgust/fear/happiness/neutral/sadness/surprise
  • blur level
  • exposure
  • noise level
  • eye makeup and lip makeup
  • accessories
  • occlusion
  • hair: bald/invisible/hair color probabilities
  • smile probability
  • face bounding rectangle

Overview

Information on downloading the images is available on the FFHQ official page. The extracted features are inside the json folder of this repository.

For example, 00000.json contains the following information:

    {
        "faceId": "866db6af-66ee-4828-b565-994f42571d2f",
        "faceRectangle": {
            "top": 42,
            "left": 32,
            "width": 63,
            "height": 63
        },
        "faceAttributes": {
            "smile": 0.003,
            "headPose": {
                "pitch": 5.4,
                "roll": 1.5,
                "yaw": 15.3
            },
            "gender": "female",
            "age": 0.0,
            "facialHair": {
                "moustache": 0.0,
                "beard": 0.0,
                "sideburns": 0.0
            },
            "glasses": "NoGlasses",
            "emotion": {
                "anger": 0.0,
                "contempt": 0.0,
                "disgust": 0.0,
                "fear": 0.0,
                "happiness": 0.003,
                "neutral": 0.993,
                "sadness": 0.003,
                "surprise": 0.0
            },
            "blur": {
                "blurLevel": "medium",
                "value": 0.55
            },
            "exposure": {
                "exposureLevel": "goodExposure",
                "value": 0.6
            },
            "noise": {
                "noiseLevel": "medium",
                "value": 0.44
            },
            "makeup": {
                "eyeMakeup": false,
                "lipMakeup": false
            },
            "accessories": [],
            "occlusion": {
                "foreheadOccluded": false,
                "eyeOccluded": false,
                "mouthOccluded": false
            },
            "hair": {
                "bald": 0.07,
                "invisible": false,
                "hairColor": [
                    {
                        "color": "brown",
                        "confidence": 0.93
                    },
                    {
                        "color": "blond",
                        "confidence": 0.8
                    },
                    {
                        "color": "black",
                        "confidence": 0.41
                    },
                    {
                        "color": "red",
                        "confidence": 0.28
                    },
                    {
                        "color": "gray",
                        "confidence": 0.22
                    },
                    {
                        "color": "other",
                        "confidence": 0.2
                    }
                ]
            }
        }
    }

Acknowledgements

We thank the authors of the FFHQ dataset. The information was extracted using a neural network. For further information, contact the author.

Licenses

The individual images were published in Flickr by their respective authors under either Creative Commons BY 2.0, Creative Commons BY-NC 2.0, Public Domain Mark 1.0, Public Domain CC0 1.0, or U.S. Government Works license. All of these licenses allow free use, redistribution, and adaptation for non-commercial purposes. However, some of them require giving appropriate credit to the original author, as well as indicating any changes that were made to the images. The license and original author of each image are indicated in the metadata.

The dataset itself (including JSON metadata, download script, and documentation) is made available under Creative Commons BY-NC-SA 4.0 license by NVIDIA Corporation. You can use, redistribute, and adapt it for non-commercial purposes, as long as you (a) give appropriate credit by citing our paper, (b) indicate any changes that you've made, and (c) distribute any derivative works under the same license.