Skip to content

Dataset · tropos-labs

Synthetic face dataset for embedding training

tropos-labs/eigen-face-dataset-256

Face image dataset with various angles, lighting, and expressions to build recognition models without privacy issues.

What's inside

According to the description, the dataset is created to train facial embedding models and includes multiple images per person with variations in pose, lighting, facial expression, and age. The set is positioned as an alternative to real photographs, avoiding privacy legal issues, ethical risks, sample imbalance, and manual labeling difficulties. Exact size, file format, and number of examples are not specified in the card.

What you can do with it

  • Training face embedding extraction models (face recognition and verification)
  • Testing the robustness of facial recognition systems to pose and lighting changes
Downloads this month
2.9M
Likes
5

How to load it

from datasets import load_dataset

ds = load_dataset("tropos-labs/eigen-face-dataset-256")