Luxand FaceSDK 8.3 — Complete Developer Guide to Face Recognition Integration

What Is Luxand FaceSDK?

Luxand FaceSDK is a cross-platform face recognition and biometric identification software development kit (SDK) developed by Luxand Inc. It provides a comprehensive C library with language wrappers for building applications that can detect faces, track facial features, recognize individuals, and perform liveness detection — all within the developer’s own application without cloud dependency.

Version 8.3 is the current release (8.2 released September 2024, 8.3 following with developer-focused improvements across mobile, .NET, WebAssembly, and cross-platform wrappers).

FaceSDK is an on-premise SDK — it runs entirely on the developer’s hardware and infrastructure. This distinguishes it from cloud API services: there is no per-call fee, no data sent to external servers, and no internet dependency during runtime. For applications handling sensitive biometric data in regulated environments (healthcare, government, enterprise security), on-premise deployment is often a requirement.

The SDK is used by developers worldwide across applications including:

Biometric access control systems replacing keycards and PINs
Time and attendance systems in workplaces, schools, and factories
Secure authentication for software login and transaction verification
Surveillance and monitoring systems with face search in video archives
Augmented reality and facial effects for entertainment applications
Photo and video search systems identifying faces in large media libraries

Platform and Language Support

FaceSDK is one of the most broadly compatible face recognition SDKs available, supporting an extensive combination of platforms and programming languages:

Supported Platforms

Windows — 32-bit and 64-bit (Windows 7 through Windows 11)
Linux — 64-bit (Ubuntu, Debian, RHEL, and other distributions)
macOS — 64-bit and Apple Silicon (M1/M2)
iOS — for iPhone and iPad applications
Android — for Android smartphone and tablet applications
Web (WebAssembly) — browser-based face recognition without plugins

Supported Programming Languages

C / C++ — native library, direct DLL/SO interface
C# (.NET) — .NET Standard 2.0 wrapper (compatible with .NET Framework, .NET Core, .NET 5+)
Python — Python wrapper with enhanced IDE compatibility
Java — for Android and desktop Java applications
Objective-C / Swift — for native iOS development
Visual Basic (VB)
Delphi / Pascal
Flutter — cross-platform mobile wrapper (iOS and Android)
React Native — cross-platform mobile wrapper (iOS and Android)

This language breadth means development teams working in any major language stack can integrate FaceSDK without switching tools or languages.

Core SDK Capabilities

Face Detection

The foundational function — locating faces in images and video frames:

Detect multiple faces simultaneously in a single image or video frame
Returns bounding rectangle (coordinates) for each detected face
Works on still images (.bmp, .jpg, .png) and live video streams
Robust to varying lighting conditions — designed to maintain accuracy across poorly lit, backlit, and high-contrast scenes
Mask-on face detection — detect faces even when subjects are wearing face masks (particularly relevant for post-pandemic environments)
Thermal image face detection — detect faces in thermal/infrared camera images, enabling operation in complete darkness
IP camera support — direct integration with RTSP streams from network cameras

Detection performance is optimized for multi-core processors — FaceSDK uses all available CPU cores to maximize throughput, enabling real-time detection in high-resolution video or across multiple simultaneous camera feeds.

70-Point Facial Feature Detection and Tracking

Beyond simple face detection, FaceSDK detects 70 precise facial landmark points, including:

Eye corners, centers, and eyelids
Eyebrow contours
Nose tip, bridge, and nostrils
Lip corners and contours (upper and lower lip)
Jawline and face contour points
Chin

These landmark coordinates enable:

Facial feature tracking in video — smooth, jitter-minimized tracking of all 70 points as the subject moves
Head pose estimation — determine face orientation (yaw, pitch, roll)
AR overlay alignment — precisely position virtual objects relative to facial features (glasses, mustaches, filters)
Photo post-processing automation — red-eye removal, skin tone adjustment, face-aware cropping
3D face model generation — build 3D head geometry from a single still image using landmark positions

حتما بخوانید: Schneider Electric OPC Factory Server 3.63 — Complete Guide for Automation Engineers

Face Recognition and Identification

The core biometric capability — determining whether two face images show the same person, and searching a database for matching identities:

1:1 Verification: Compare two face images or templates and receive a similarity score — used for authentication (verify that the person at the camera matches the enrolled identity).

1:N Identification: Search a database of enrolled face templates and return the closest match — used for surveillance, time-and-attendance, and access control where the subject’s identity is determined from their face alone.

Face template extraction: Extract a compact numerical face template (feature vector) from a detected face — templates are small (a few kilobytes), enabling efficient database storage and fast comparison. Templates can be stored and compared without retaining the original face image, which simplifies privacy compliance.

Recognition characteristics:

Stable recognition robust to lighting variation, facial expression changes, and moderate aging
Low false positive rate — designed for security applications where incorrect identifications are costly
Optimized for real-world operating conditions — not just controlled lab environments

Tracker API — Self-Learning Video Identification

The Tracker API is designed specifically for continuous video stream monitoring:

Automatic face discovery — as new faces appear in the video stream, the Tracker detects and tags them automatically
Tag and name assignment — once a face is tagged (manually or through database match), the Tracker continues recognizing that person across subsequent frames, even through re-entries after leaving the frame
Self-learning — the Tracker accumulates face samples over time, improving its ability to recognize previously seen individuals even under varying conditions
Multi-camera support — thread-safe design enables simultaneous processing of multiple video streams on the same server

The Tracker API dramatically simplifies building attendance and surveillance systems — developers call a small set of functions rather than managing frame-by-frame detection and matching logic.

Liveness Detection — Anti-Spoofing (iBeta Certified)

Liveness detection distinguishes a real, present human face from a spoofing attack (photograph, printed mask, video replay, 3D mask):

iBeta Certified: Luxand’s Liveness add-on has passed Level 1 iBeta Presentation Attack Detection (PAD) conformance testing per ISO/IEC 30107-3. iBeta is accredited by NIST/NVLAP (Lab Code: 200962). During testing, approximately 1,000 presentation attacks using authentic biometric samples were conducted — none succeeded, resulting in an APCER (Attack Presentation Classification Error Rate) of 0%.

Detection modes:

Passive liveness — analyzes a single frame or short video clip without requiring user action (blink, turn head). Examines texture, depth cues, and presentation artifact signatures
Active liveness — prompts the user to perform a specific action (blink, smile, turn head) and verifies the response is from a live person

Platform availability: Currently available for Android and Windows/x64; additional platforms planned.

For applications in banking, healthcare, government ID verification, and high-security access control, liveness detection is essential to prevent spoofing attacks.

Attribute Recognition

Beyond identity, FaceSDK recognizes additional facial attributes:

Gender — male/female classification
Age estimation — estimated age range
Expression recognition:
- Smile detection (present/absent)
- Eye state (open/closed)
These attributes can be used for demographic analysis, user experience personalization, or as additional authentication signals

Key APIs

Image Management API

FaceSDK includes a complete image handling layer:

Load images from files (.bmp, .jpg, .png) or memory buffers
Resize, rotate, crop, flip images
Pixel-level editing
Save to the same formats
Useful for preprocessing pipeline steps before face detection

Face Detection API (Basic)

// C# example
FSDK.TFacePosition facePos;
FSDK.DetectFace(imageHandle, ref facePos);
// Returns bounding box coordinates

Feature Detection API

// Get 70 facial feature coordinates
FSDK.TPoint[] features = new FSDK.TPoint[FSDK.FSDK_FACIAL_FEATURE_COUNT];
FSDK.DetectFacialFeatures(imageHandle, ref features);

Face Template API

// Extract face template for database storage/comparison
byte[] template = new byte[FSDK.TemplateSize];
FSDK.GetFaceTemplate(imageHandle, ref template);

// Compare two templates
float similarity;
FSDK.MatchFaces(template1, template2, ref similarity);
// similarity: 0.0 (no match) to 1.0 (identical)

Tracker API

// Initialize tracker
long trackerHandle;
FSDK.CreateTracker(ref trackerHandle);

// Process video frames
long[] faceIDs;
FSDK.FeedFrame(trackerHandle, 0, imageHandle, ref faceCount, ref faceIDs, ...);

// Get recognized name for a face ID
string name;
FSDK.GetName(trackerHandle, faceIDs[i], ref name, 256);

Application Use Cases

Access Control System

Replace physical keycards with face recognition at entry points. Enrolled employees are recognized at the door; unrecognized faces trigger an alert. The Tracker API handles continuous video monitoring, and liveness detection prevents photo spoofing. Integration with door controller hardware is through the application layer — FaceSDK provides only the face recognition engine.

حتما بخوانید: archelios CALC — Complete Guide for PV Electrical Sizing and Compliance (2026)

Time and Attendance

Employees clock in by facing a camera. FaceSDK identifies each person, records attendance with timestamp, and generates daily reports. No badge swipes, no PIN entry — frictionless and forgery-resistant. The self-learning Tracker API maintains recognition quality even as employees’ appearance changes over time.

Biometric Login for Desktop/Mobile Apps

Replace passwords with face authentication. The user looks at the webcam or front camera; FaceSDK verifies their identity against the enrolled template. Liveness detection prevents authentication with a photo. Integration in C# for Windows desktop, Swift for iOS, or Java for Android — all using the same FaceSDK recognition engine.

Photo Library Face Search

A media management application needs to find all photos containing a specific person. FaceSDK processes the photo library, extracts templates from all detected faces, and enables fast similarity-based search — returning all images where the query person appears. Handles databases of millions of images with GPU acceleration where available.

Surveillance and Security Monitoring

Match faces appearing in video footage against a database of persons of interest. The Tracker API processes live camera feeds; detected faces are matched against the watch list in real time. Alerts are generated when a match above the confidence threshold is found.

AR Facial Filters

Apply virtual accessories (glasses, hats, masks, effects) precisely aligned to facial features. The 70-point landmark detection provides the anchor points for overlay positioning; tracking ensures overlays follow face movement smoothly through video frames. Used in entertainment apps, virtual try-on for eyewear and cosmetics, and social media filter tools.

FaceSDK vs. Competing Face Recognition SDKs

Feature	Luxand FaceSDK 8.3	Microsoft Azure Face API	AWS Rekognition	OpenCV + dlib (free)
On-premise deployment	✅	❌ Cloud only	❌ Cloud only	✅
No per-call fee	✅	❌ Pay per call	❌ Pay per call	✅
Internet required at runtime	❌	✅	✅	❌
iOS/Android/Web support	✅ All	Via API call	Via API call	Limited
C#/Python/Java/Delphi	✅ All	Limited	Limited	Python/C++ only
70-point landmark detection	✅	✅	Limited	✅ (68-point)
Liveness detection (iBeta PAD)	✅ Certified	✅	✅	❌
Mask detection	✅	✅	✅	❌
Thermal camera support	✅	❌	❌	Limited
Tracker API (video)	✅	❌	❌	Manual implementation
Self-learning video tracking	✅	❌	❌	❌
WebAssembly (browser)	✅ 8.1+	Via API	Via API	Limited
Flutter/React Native	✅ 8.1+	Limited	Limited	❌
.NET Standard 2.0 wrapper	✅	Via HTTP client	Via HTTP client	❌
GDPR on-premise data control	✅ Full	Limited	Limited	✅
Pricing model	One-time license	Per API call	Per API call	Free/Open source

Choose FaceSDK when: You need on-premise deployment (no cloud dependency), are building in Delphi/VB/.NET/native C++ environments, need thermal camera support, or want the Tracker API for automatic video identification without manual frame management. Also preferred when GDPR or data sovereignty requires biometric data to remain entirely within your infrastructure.

Choose cloud APIs (Azure, AWS) when: You prefer operational pricing over license cost, need zero infrastructure maintenance, and can accept cloud data transmission requirements.

Choose OpenCV/dlib when: Cost is the primary constraint, your team has computer vision expertise to build surrounding infrastructure, and you don’t need commercial support or liveness detection certification.

حتما بخوانید: Geneious Prime 2026 — Bioinformatics & Sequence Analysis Software | Complete Guide

Licensing Options

FaceSDK is available in several license tiers depending on deployment scale:

Developer License — for individual developers building and testing applications (not for deployment)

Single Application License — for one commercial application deployed on end-user machines or servers

Site License — for deployment across a defined number of installations at a specific site or organization

OEM License — for embedding FaceSDK in commercial software products that are resold to third parties

Custom/Enterprise — for large-scale deployments; contact Luxand for custom pricing

Licenses are perpetual (not subscription) — purchase once, deploy indefinitely on the licensed number of installations. Software updates during the license period vary by tier.

A free trial/evaluation version is available from Luxand’s website — fully functional for evaluation but requires a commercial license for production deployment.

System Requirements

Component	Requirement
OS	Windows 7+, Linux (64-bit), macOS 10.13+, iOS 11+, Android 5.0+
CPU	Multi-core x86/x64 or ARM (Apple Silicon, ARM64)
RAM	512 MB minimum; 2+ GB recommended for large databases
GPU	Optional — CPU-based processing as standard
Storage	~1 GB for SDK installation
Network	Not required at runtime (on-premise)

Frequently Asked Questions

Does FaceSDK require cloud connectivity during recognition? No. FaceSDK is entirely on-premise. After installation and license activation, no internet connection is required during runtime. All face detection, recognition, and template matching occurs locally on the deployment hardware.

How many faces can the database hold for 1:N identification? FaceSDK itself does not impose a hard database limit — the practical limit depends on available RAM and required response time. Template comparison is very fast (each template comparison is microsecond-scale on modern hardware), making databases of tens of thousands of individuals feasible on standard server hardware. For millions of identities, database indexing and partitioning strategies are needed at the application layer.

Is FaceSDK GDPR compliant? FaceSDK itself is a library — GDPR compliance is the responsibility of the application built with it. The on-premise nature of FaceSDK is a significant advantage for GDPR: biometric data (face templates) never leaves your infrastructure. Templates can be stored separately from identity information, and FaceSDK does not require storing original face images after template extraction.

What languages are supported for mobile development? For iOS: Objective-C, Swift, and Flutter wrapper. For Android: Java, and both Flutter and React Native wrappers. All mobile wrappers were updated/added in FaceSDK 8.1.

Can FaceSDK process video from IP cameras directly? Yes. FaceSDK supports IP camera RTSP streams for both detection and the Tracker API. The application connects to the camera stream and passes frames to FaceSDK — Luxand provides documentation and samples for this integration.

What is the difference between FaceSDK and Luxand.cloud? FaceSDK is an on-premise SDK licensed for deployment on your own servers and devices. Luxand.cloud is a REST API cloud service offering the same recognition capabilities on a per-call pricing model. For data-sensitive or offline applications, FaceSDK is the appropriate choice.

Summary

Luxand FaceSDK 8.3 is one of the most complete and broadly compatible face recognition SDKs available for on-premise deployment. Its combination of extensive language and platform support (C++, C#, Python, Java, Swift, Delphi, Flutter, React Native across Windows, Linux, macOS, iOS, Android, and WebAssembly), production-grade recognition quality, iBeta-certified liveness detection, Tracker API for autonomous video monitoring, and thermal camera support makes it a strong choice for development teams building biometric applications across security, enterprise, entertainment, and identity verification domains.

For licensing assistance or integration support, contact our team via Telegram: t.me/DoCrackMe