FaceTec General Technical Info
The following tables contain detailed information about the FaceTec 3D FaceScans, 3D FaceMaps & SDKs
General Payload Sizes
| 3D FaceScan Size* | ~430 KB - 515 KB (in transit; includes Liveness Data & 3D FaceMap for Matching) |
|---|---|
| 3D FaceMap Size** | ~80 KB - 300 KB (does not include Liveness Data) |
| 3D FaceScan/Map Encryption | Asymmetric RSA & Randomized AES-256 |
| Audit Trail Image Size | 640px tall and ~50KB |
* 3D FaceScan Size is dependent on the Device and Camera.
** The initial size of a 3D FaceMap is around 80 KB. Since 3D FaceMaps are adaptive, they can continue to grow up to ~300 KB as new information is retrieved from Subsequent 3D:3D Matches of the User.
Device SDK Details
| Minimum iOS/Android/Browser Version | |
|---|---|
| Minimum Camera Requirements | 0.3 Megapixel (VGA) -- All Known Cameras |
| Minimum CPU | iOS/Android - All armv7 and arm64. |
| Minimum GPU | No requirements |
| Minimum Memory | No known device incompatibility due to memory |
| Minimum Disk | No known device incompatibility due to disk |
| iOS & Android: MB Added to Your App | |
| Browser / Mobile Browser: MB Added to Your Webpage |
FaceTec Server Details
| Operating System (FaceTec Server Core) | Most Modern Linux Distros (e.g. Debian, CentOS, RedHat, Ubuntu, and Amazon Linux) |
|---|---|
| Windows Supported | Yes, via Docker (should only be used in early development and NOT in a production environment) |
| Mac Supported* | Yes, via Docker (should only be used in early development and NOT in a production environment) |
| Linux Supported | Yes, via Docker |
| FaceTec REST Sample Code Language | Java |
| Oldest iOS/Android/Browser Versions | See Compatibility Page |
* The FaceTec Server has limited compatibility with Mac Silicon. While the FaceTec Server can initialize on a Mac Silicon, subsequent API calls will throw standard FaceTec exceptions. For more information, see FAQ - Is the FaceTec Server supported on Mac Silicon and other ARM CPUs?
Server Hardware (Minimum Recommendations)
| Server CPU | Number of CPUs: 16 CPUs strongly recommended* |
|---|---|
| Server Memory (Liveness/Enrollment/3D Matching/Photo ID Match) | Number of CPUs * 2.5 GB of memory/RAM (e.g. 8 CPU needs 20 GB RAM; 16 CPU needs 40 GB RAM) |
| Server Memory (Search) | Number of CPUs * 2.5 GB of memory/RAM + 10GB per 1 Million Enrollments |
| Server Disk Space (Liveness Check Only) | 20GB |
| Server Disk Space (3D Enrollment/3D:3D Re-Verification/Photo ID Matching) | 20GB** |
| Server Disk Space (Search) | 20GB + 10GB per 1 Million Enrollments |
| Server Networking | GB+ Internal Adapters/Networks Recommended |
| Virtualization | Supported. Please note some VM configurations do not allow leveraging CPU hardware acceleration and should be tested. |
| GPU | No GPU requirements. |
| AWS EC2 Instance Baseline Hardware | Intel Processors: c7i.2xlarge or c7i.4xlarge. See AWS EC2 C7i Instances Documentation for more details. |
* While 16 CPUs is strongly recommended, an 8-CPU Integration may be possible depending on the configuration and your team's expected usage.
** Assumes 3D FaceMaps (Enrollments) are stored in a Customer's database or external filesystem.
FaceTec strongly advises using the Recommended Hardware to prevent incurring problems in your team's Development and Production Environments, such as Server Initialization-Related Behaviors, Degraded Performance, and Crashes/Exceptions, all of which can lead to Service Disruption.
Databases
| FaceTec Server | In-Memory Stub Database, MongoDB, DocumentDB, S3, PostgreSQL, MySQL, MSSQL, Oracle DB, and more. |
|---|---|
| FaceTec Dashboard SDK | MongoDB, DocumentDB, or PostgreSQL is required. |
| AWS DB/Storage Support | S3, DynamoDB, DocumentDB, RDS, Aurora, and more. |
| Google Cloud DB/Storage Support | SQL, Bigtable, and more. |
| Azure DB/Storage Support | SQL, Cosmos, and more. |
| Other DB/Storage Support | MongoDB, PostgreSQL, MySQL, and more. |
Data Types & Storage
The Database is owned and managed by Your Team. You and Your Team have 100% control over Data Storage and Deletion.
The FaceTec Server's Prebuilt Configuration stores FaceMaps, Images, and Other Metadata by default in order to support the Server Dashboard GUI. For more information on creating a Data Storage Strategy, see Reviewing & Altering Data Storage in Your Architecture.
The FaceTec Server's Recompiled Configuration stores data via Your Team's Code. Data Storage and Retrieval, along with any Metadata that is stored, is 100% configurable and designed by You and Your Team. The REST Sample Code is provided as an Initial Example and should be modified to suit Your Team's Business Requirements.
| Data Type | Description | Notes |
|---|---|---|
| 3D FaceMaps | Encrypted Biometric from Liveness Check. | Required to perform Liveness Checks, Enrollment, Re-Verification, and Photo ID Matching. Storage is required for Enrollment, Re-Verification, and Photo ID Matching. |
| 2D Audit Trail Images | Image Data from User Session. | Required for certain Security Checks. |
| ID Scans | Encrypted data from Photo ID Match Session. | Allows 1/500,000 FAR 3D FaceMap to Photo ID Matching with ID Anti-Spoofing. |
| ID Scan Images | Front and Back Images of User ID. | Can be used to perform additional ID Verification. |
| Additional Metadata | e.g. FaceTec Version, IP Addresses, Device Models. | Required for Server Dashboard display. 100% configurable. |
Server Benchmarks & Estimated Costs
| Baseline Cluster Specs | Intel CPU Baseline: 4 servers, each running 8-core or 16-core Intel Xeon Platinum 8000 Series CPUs |
|---|---|
| Liveness Check Throughput (Baseline Cluster)* | ~2 Liveness Checks per second per server |
| Match 3D:2D IDScan Throughput (Baseline Cluster)* | ~10 - 20 scans per minute per server |
| IDScan Only Throughput (Baseline Cluster)* | ~10 - 20 scans per minute per server |
| AWS Cluster Machine Type | CPU (Intel): c7i.2xlarge or c7i.4xlarge |
| Azure Cluster Machine Type | Standard_F16s_v2 |
| Google Cloud Cluster Machine Type | n2-standard-8 or n2-standard-16 |
| Compute Cost (Baseline Cluster) | AWS or Google: $5400-$6200 annually |
While not mandatory, Server Clusters are highly recommended and allow for higher availability, proper load balancing, and system scalability in your Production Environment. To ensure optimal scalability and availability, FaceTec recommends Stress/Load Testing to calculate your Server Clusters' max throughput, while using Horizontal Scaling to handle peak loads and high-traffic periods. See FaceTec Server Stress Testing Guide for more details.
Cloud Providers & Misc. Technologies
| Custom Windows Servers | Supported, via Docker |
|---|---|
| Custom Linux Servers | Supported, via Docker |
| AWS EC2 | Supported |
| AWS EB | Supported |
| AWS Lambda | Not Supported due to Lambda app size restriction. |
| AWS Fargate | Supported* |
| Kubernetes | Supported* |
| Google Compute Engine | Supported* |
| Azure VMs | Supported* |
| Azure Kubernetes | Supported* |
* Known deployments exist (Partner, Customer, etc.) with this technology.