Live-recorded attacks captured via low-quality web-cameras. The dataset solves the tasks of liveness detection and security

Low Quality Webcam Live Attacks

Ability of a machine to interpret, analyze, and understand visual data
Computer Vision
Process of identifying or verify a person identity using facial features
Facial Recognition
The ability to distinguish whether biometric data is being captured from a live or fake source
Liveness Detection
Techniques used to prevent fraudulent attempts to deceive an identifying system
1 000
8 weeks
project duration
Live videos of people collected from crowdsourcing platforms
Different video resolutions: QVGA (320 x 240p), QQVGA (160 x 120 p), QCIF (176 x 144 p)
Types of light: indoor daytime and evening lighting
Unique attack identifier
Identifier of the user recording the attack
User's age
User's gender
Metadata is represented in the file_info.csv. Each attack instance is accompanied by the following details:
User's country of origin
Attack resolution
The model of the webcam
Data collection of Live selfies and videos of people from webcams with a resolution from Full HD to 4K with a volume and metadata on demand
Data Labelling: Bounding Box and classification for selfies and videos
Tell us about your project!
See other datasets