All-Age-Faces Dataset

Category: Themed Directories
Type: Public machine learning image dataset
Source: GitHub – All-Age-Faces Dataset

Overview

The All-Age-Faces (AAF) Dataset is a curated collection of 13,322 mostly Asian face images covering ages approximately 2 to 80. It is designed for machine learning research tasks such as age prediction and gender classification, and can also support face analysis across different age ranges.

Features

• Size and Composition

  • 13,322 face images in total
  • Mostly Asian subjects
  • Age range: 2 to ~80 years
  • Gender distribution:
    • 7,381 female images
    • 5,941 male images
  • Each image contains a different individual (no identity repetition)

• Data Organization

  • original images folder: original face images
  • key points folder: facial landmark annotations
  • aligned faces folder: aligned face images based on landmarks
  • example folder: example of facial landmark distribution
  • image sets folder: train/validation split annotation files

• File Naming Convention

  • Image files named as: %05dA%02d.jpg
    • person_id: 5-digit serial number (00000–13321)
    • age: 2-digit age value
  • Gender by serial number range:
    • 00000–07380: female
    • 07381–13321: male

• Annotations & Splits

  • Each image is annotated with person ID, age, and gender
  • Standard random split into training and validation sets for fair comparison
  • Annotation line format in image sets:
    • "%05dA%02d %d\n", person_id, age, gender
    • gender = 0 for female, 1 for male

• Intended Use Cases

  • Age prediction / age estimation
  • Gender classification
  • Face analysis across different ages
  • Research on facial landmarks and alignment-based methods

• Access & Downloads

  • Baidu link: https://pan.baidu.com/s/1WtHQsb73rLa-cZpBLi2dtg
  • Dropbox link: https://www.dropbox.com/s/a0lj1ddd54ns8qy/All-Age-Faces%20Dataset.zip?dl=0

• Contact

  • Maintainer: Jingchun Cheng
    Email: chengjingchun [at] gmail [dot] com

• Citation

  • If used in research, cite:

    @article{cheng2019exploiting,
      title={Exploiting effective facial patches for robust gender recognition},
      author={Cheng, Jingchun and Li, Yali and Wang, Jilong and Yu, Le and Wang, Shengjin},
      journal={Tsinghua Science and Technology},
      volume={24},
      number={3},
      pages={333--345},
      year={2019},
      publisher={TUP}
    }
    

Pricing

• The content does not specify any pricing; the dataset is presented as a public research dataset (check the repository for any license terms or usage conditions).