dwancin/facial-embeds

Facial Embeds

A dataset of facial embeddings, ideal for tasks in image classification and face recognition.

Overview

Facial Embeds is a comprehensive dataset featuring over 24.9k distinct facial embeddings. Designed for enthusiasts and practitioners in computer vision, it supports a variety of applications from image classification to face recognition.

Details

• Creator: Daniel Ialcin Misser Westergaard
• License: MIT
• Repository: dwancin/facial-embeds
• Demo: dwancin/face-recognition

Features

• Supported Tasks: Zero-shot image classification, Face recognition, Face verification, Face clustering
• Tags: Computer vision, Face recognition, Face detection, Facenet, Machine learning

Data Structure

Each record in the dataset represents an image, organized as follows:

• id: Unique identifier.
• file_name: Path to image ('images/{file_name}').
• text: Name or label associated with the image.
• embeddings: Numerical facial embedding vector.

Example Entry

ID	File Name	Text	Embeddings
1	images/1.jpg	Person Name	[0.1, 0.2, ..., 0.0]
2	images/2.jpg	Person Name	[0.2, 0.3, ..., 0.0]
3	images/3.jpg	Person Name	[0.3, 0.4, ..., 0.0]
...	...	...	...