metadata

license: apache-2.0
task_categories:
  - image-classification
size_categories:
  - 10K<n<100K
tags:
  - Anime
  - Cartoon
  - Realistic
  - Sketch
  - Portrait
  - art

Multilabel-Portrait-18K

Multilabel-Portrait-18K is a multi-label portrait classification dataset designed to analyze and categorize different styles of portrait images. It supports classification into the following four portrait types:

0 — Anime Portrait
1 — Cartoon Portrait
2 — Real Portrait
3 — Sketch Portrait

This dataset is ideal for training and evaluating machine learning models in the domain of portrait-style classification. The goal is to enable accurate recognition of artistic and real-world portraits for applications such as image generation, enhancement, style transfer, and content moderation.

Use Cases

Multi-label classification for style recognition
Pretraining or fine-tuning portrait classifiers
Improving filters and sorting in creative AI applications
Enhancing deepfake detection via portrait-style understanding
Style-transfer or portrait enhancement tools

Dataset Details

Total Samples: 18,000 portrait images
Labels: Multi-label format (each image may have more than one label)
Label Schema:
- 0: Anime Portrait [4,444]
- 1: Cartoon Portrait [4,444]
- 2: Real Portrait [4,444]
- 3: Sketch Portrait [4,444]

Format

The dataset is typically provided in either:

A directory structure grouped by label
Or a .csv / .json file containing filename and labels fields

Example (.csv):

filename,label
portrait_001.jpg,"[0, 3]"
portrait_002.jpg,"[2]"
portrait_003.jpg,"[1, 2]"