SECURE, IN-MOTION 3D FACE IDENTIFICATION & ACCESS CONTROL
FOR HIGH VOLUME LOCATIONS
Most face recognition systems falter with real-world conditions found in high-volume access control situations — lighting is too bright or dark, people move quickly without stopping to position themselves for the camera.
3DSafe’s unique technology:
Works with no light, back light, side light, & direct sunlight.
Detects all skin tones & races.
Learns appearance changes like beards, glasses, & face occlusions.
Doesn’t require careful face alignment.
Rejects spoofing by masks, photos, & videos.
Stops tailgating by multiple users.
what makes 3dsafe better?
SensibleVision has designed 3DSafe to eliminate all the problems common with most traditional access control solutions. Lighting issues and time-consuming face alignment are things of the past. And we help you leverage your existing 2D enrollment database by aligning that 2D data within True 3D space for increased accuracy.
| 3DSafe | Others | |
|---|---|---|
| Works in-motion without people slowing down | ✔ | ✘ |
| Works off-axis without careful face alignment | ✔ | ✘ |
| Prevents tailgating by recognizing multiple people continuously | ✔ | ✘ |
| Uses and improves existing 2D enrollment databases | ✔ | ✘ |
| Works in all lighting (including no light & sunlight) | ✔ | ✘ |
| Works for all skin tones & races | ✔ | ✘ |
| Immune to spoofing with masks, photos, & videos | ✔ | ✘ |
| Uses 2X more data so beards and makeup aren't an issue | ✔ | ✘ |
3DSAFE
UNIQUE TECHNOLOGIES
Simultaneous multi-camera support
Continuous Security
Simultaneous anti-spoofing & recognition
True 3D recognition + simultaneous 2D recognition
SPECIFICATIONS
False Rejection Rate: Approaches 0% with 1-3 uses
FAR: Better than 1 in 1,000,000 without occlusions
Supports 1 to Many configurations (1:N) & databases of 10,000+
With beards and occlusions:
1:1 FAR of better than 1 in 1,000,000 with beards typical in the Middle East/Europe. 1:N better than 1 in 800,000
With both a beard and normal-sized sunglasses, we would expect a 1:1 FAR of better than 1 in 250,000. 1:N better than 1 in 200,000.
1:1 FAR of better than 1 in 1,000,000 with makeup. 1:N better than 1 in 800,000/
Yaw and roll: up to +/- 45 degrees supported for off-axis recognition (optional +/- 60 degrees)
Speed: Typically less than 70ms (Detection, Extraction, Template Creation, & Matching)
1,0000 users on an i7, 16GB RAM, and SSD
Larger databases can be supported with similar performance and suitable hardware
On-premises or cloud operation
Multiple camera support: for the widest field of views to scan large areas while recognizing people of different heights
Works with the leading 3D camera solutions and technologies at all price points including:
Time of Flight (ToF)
Structured Light (SL)
Active Stereo (AS)