Skip to content
intermediateimage

CelebA: Large-Scale Face Attributes Dataset

CelebA is a face dataset of 202,599 celebrity images, each labeled with 40 binary attributes, identity, and 5 landmark points (Liu et al., ICCV 2015). This Potato config reproduces the 40-attribute multiselect task.

About this dataset

CelebFaces Attributes (CelebA) was built by Ziwei Liu, Ping Luo, Xiaogang Wang, and Xiaoou Tang at the Multimedia Lab, Chinese University of Hong Kong, and released with the paper "Deep Learning Face Attributes in the Wild" at ICCV 2015.

The dataset holds 202,599 face images covering 10,177 celebrity identities, with wide pose variation and background clutter. A standard split assigns images 1-162,770 to training, 162,771-182,637 to validation, and 182,638-202,599 to testing.

Each image carries 40 binary attribute labels (for example hair color, smiling, eyeglasses, wearing hat), an identity label, and 5 landmark locations marking the eyes, nose, and mouth corners. The data supports face attribute recognition, face detection, and landmark localization. In-the-wild and aligned-and-cropped image variants are both provided.

The Potato config below reproduces the 40-attribute labeling task: an annotator views one face image and checks every attribute that applies from the 40 binary categories using a multiselect scheme.

Face images
202,599
Identities
10,177
Binary attributes
40 per image
Landmark points
5 per image
Venue
ICCV 2015
Train / val / test
162,770 / 19,867 / 19,962
Labels:outdoornatureurbanpeopleanimal+−

Configuration Fileconfig.yaml

This Potato config reproduces the annotation task. Save it as config.yaml and run potato start config.yaml to try it.

yaml
# CelebA Face Attributes Classification Configuration
# Based on Liu et al., ICCV 2015

annotation_task_name: "CelebA Face Attributes"
task_dir: "."

data_files:
  - "sample-data.json"

item_properties:
  id_key: "id"
  text_key: "image_url"
  context_key: "context"

user_config:
  allow_all_users: true

annotation_schemes:
  - annotation_type: "multiselect"
    name: "hair_attributes"
    description: "Select hair attributes"
    labels:
      - name: "bald"
        tooltip: "Little or no hair"
      - name: "bangs"
        tooltip: "Has bangs/fringe"
      - name: "black_hair"
        tooltip: "Black hair color"
      - name: "blond_hair"
        tooltip: "Blonde hair color"
      - name: "brown_hair"
        tooltip: "Brown hair color"
      - name: "gray_hair"
        tooltip: "Gray or white hair"
      - name: "straight_hair"
        tooltip: "Straight hair"
      - name: "wavy_hair"
        tooltip: "Wavy hair"
      - name: "receding_hairline"
        tooltip: "Receding hairline"

  - annotation_type: "multiselect"
    name: "facial_features"
    description: "Select facial features"
    labels:
      - name: "arched_eyebrows"
        tooltip: "Arched eyebrows"
      - name: "bushy_eyebrows"
        tooltip: "Bushy eyebrows"
      - name: "big_lips"
        tooltip: "Full/big lips"
      - name: "big_nose"
        tooltip: "Big nose"
      - name: "pointy_nose"
        tooltip: "Pointy nose"
      - name: "high_cheekbones"
        tooltip: "High cheekbones"
      - name: "narrow_eyes"
        tooltip: "Narrow eyes"
      - name: "bags_under_eyes"
        tooltip: "Bags under eyes"
      - name: "double_chin"
        tooltip: "Double chin"

  - annotation_type: "multiselect"
    name: "facial_hair"
    description: "Select facial hair attributes"
    labels:
      - name: "5_o_clock_shadow"
        tooltip: "Light stubble"
      - name: "goatee"
        tooltip: "Goatee"
      - name: "mustache"
        tooltip: "Mustache"
      - name: "no_beard"
        tooltip: "Clean shaven"
      - name: "sideburns"
        tooltip: "Sideburns"

  - annotation_type: "multiselect"
    name: "accessories"
    description: "Select accessories"
    labels:
      - name: "eyeglasses"
        tooltip: "Wearing eyeglasses"
      - name: "wearing_hat"
        tooltip: "Wearing a hat"
      - name: "wearing_earrings"
        tooltip: "Wearing earrings"
      - name: "wearing_necktie"
        tooltip: "Wearing necktie"
      - name: "wearing_lipstick"
        tooltip: "Wearing lipstick"
      - name: "heavy_makeup"
        tooltip: "Heavy makeup"

  - annotation_type: "multiselect"
    name: "demographics"
    description: "Select demographic attributes"
    labels:
      - name: "male"
        tooltip: "Male presenting"
      - name: "young"
        tooltip: "Young (under ~40)"
      - name: "attractive"
        tooltip: "Conventionally attractive"
      - name: "smiling"
        tooltip: "Smiling expression"

interface_config:
  item_display_format: "<img src='{{text}}' style='max-width:100%; max-height:500px;'/><br/><small>{{context}}</small>"

output_annotation_format: "json"
output_annotation_dir: "annotations"

Sample Datasample-data.json

json
[
  {
    "id": "celeba_001",
    "image_url": "https://upload.wikimedia.org/wikipedia/commons/thumb/c/c2/Aiga_toiletsq_men.svg/800px-Aiga_toiletsq_men.svg.png",
    "context": "Annotate face attributes: hair, facial features, facial hair, accessories, and demographics."
  },
  {
    "id": "celeba_002",
    "image_url": "https://upload.wikimedia.org/wikipedia/commons/thumb/4/4d/Cat_November_2010-1a.jpg/1200px-Cat_November_2010-1a.jpg",
    "context": "Select all applicable face attributes from the CelebA attribute list."
  }
]

// ... and 1 more items

Get This Design

View on GitHub

Clone or download from the repository

Quick start:

git clone https://github.com/davidjurgens/potato-showcase.git
cd potato-showcase/image/specialized/celeba
potato start config.yaml

Dataset & paper

Liu et al., ICCV 2015

Official dataset ↗Read the paper ↗

Citation (BibTeX)

bibtex
@inproceedings{liu2015deep,
    title = "Deep Learning Face Attributes in the Wild",
    author = "Liu, Ziwei  and Luo, Ping  and Wang, Xiaogang  and Tang, Xiaoou",
    booktitle = "Proceedings of the IEEE International Conference on Computer Vision",
    pages = "3730--3738",
    year = "2015",
    url = "https://openaccess.thecvf.com/content_iccv_2015/html/Liu_Deep_Learning_Face_ICCV_2015_paper.html"
}

Details

Annotation Types

multiselect

Domain

Computer VisionFace Analysis

Use Cases

Face Attribute ClassificationFace DetectionDemographic Analysis

Tags

celebafaceattributescelebrityiccv2015

Found an issue or want to improve this design?

Open an Issue

Related Designs

ADE20K: Scene Parsing and Semantic Segmentation Dataset

ADE20K is a scene parsing dataset from MIT CSAIL (Zhou et al., CVPR 2017) with dense pixel labels for objects, stuff, and parts. This Potato config adapts it to an image-level category-tagging task over a curated subset of common ADE20K categories.

multiselectradio

BDD100K: Diverse Driving Dataset for Autonomous Driving

BDD100K is a 100,000-video driving dataset from UC Berkeley (Yu et al., CVPR 2020) spanning detection, segmentation, lane marking, and tracking. This Potato config reproduces its driving-scene labeling tasks.

multiselectradio

Cityscapes Instance Segmentation

Urban scene understanding with instance-level semantic labeling (Cordts et al., CVPR 2016). Annotate street scenes with pixel-level labels for 30 classes across vehicles, humans, construction, and nature.

multiselect