Skip to content

Dataset · gofixyourself

20,000 portraits with facial feature and background masks

gofixyourself/EasyPortrait

A large dataset of 20,000 RGB portraits with annotations for person segmentation and facial elements.

What's inside

The dataset is about 26 GB in size and includes 20,000 color images, of which approximately 17,500 have FullHD resolution. The pictures are provided with high-quality annotations for portrait segmentation and parsing individual facial areas. The split sizes in the metadata fall in the range of 10,000 to 100,000 records.

What you can do with it

  • Training models for background blurring and replacement in video conferences
  • Creating algorithms for skin retouching, teeth whitening, and eye color modification
Tasks
Image segmentation
Size
10K–100K rows
Downloads this month
1.3K
Likes
8

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("gofixyourself/EasyPortrait")