TrainingDataPro/biometric-attacks-in-different-lighting-conditions

Dataset Overview

Dataset Name

• Biometric Attack Dataset - Different Lighting Conditions Dataset

Dataset Description

• This dataset includes personal and attack videos showing photos captured under various lighting conditions (darkroom, daylight, lightroom, and nightlight) and locations (indoor, outdoor). Each video is approximately 20 seconds long.

Video Types

• darkroom_photo - Personal photos taken in a darkroom, displayed on a computer and captured with a mobile phone
• daylight_photo - Personal photos taken in daylight, displayed on a computer and captured with a mobile phone
• lightroom_photo - Personal photos taken in a lightroom, displayed on a computer and captured with a mobile phone
• nightlight_photo - Personal photos taken under night lighting, displayed on a computer and captured with a mobile phone
• darkroom_video - Video of a person’s head moving left, right, up, and down in a darkroom
• daylight_video - Video of a person’s head moving left, right, up, and down in daylight
• lightroom_video - Video of a person’s head moving left, right, up, and down in a lightroom
• nightlight_video - Video of a person’s head moving left, right, up, and down under night lighting
• outline - Video of a person wearing a printed 2D mask
• mask - Video of a person wearing a printed 2D mask with eye cutouts
• monitor_video - Video of a personal video played on a computer and captured with a mobile phone

Dataset Contents

• files - Contains original videos and attack videos
• dataset_info.csv - Contains information about the videos in the dataset

Dataset File Information

• file: Video link
• type: Video type

License

• cc-by-nc-nd-4.0

Task Category

• video-classification

Language

• en

Tags

• code
• legal
• finance

Commercial Use

• The full version includes 7,296 videos and can be purchased via TrainingData

Dataset Purpose

• This dataset is used to learn and extract facial features through deep neural networks and advanced biometric technologies to prevent spoofing attacks. Research shows that the technology can effectively prevent unauthorized access and detect unseen spoofing cues.