Local CCTV AI Monitoring.
Zero Cloud. Zero Latency.
Hardware-accelerated on-device inference using LiteRT GPU, NCNN Vulkan, and ONNX Runtime NNAPI. Connect IP cameras, RTSP streams, device cameras, NVRs, and DVRs with 100% local privacy and continuous foreground monitoring.
Supports Android 8.0+ (Oreo, API 26+) • Package: com.gyn.lite • Zero account or cloud subscription needed.
Engineered for High-Performance Edge Surveillance
GYN lite bypasses expensive cloud NVR subscriptions by bringing enterprise-grade computer vision models directly to Android edge hardware.
Triple Hardware-Accelerated Engines
Execute local models natively on Android devices. Google LiteRT unleashes GPU shaders with CPU fallback, Tencent NCNN accelerates via Vulkan compute & ARM NEON assembly, and Microsoft ONNX Runtime powers end-to-end models with Android NNAPI acceleration.
Multi-Source CCTV & Stream Ingest
Seamlessly monitor high-definition RTSP streams, IP security cameras, connected NVRs, DVR systems, or Android's integrated CameraX sensor with continuous foreground service execution.
100% Local Privacy & Air-Gapped
Your CCTV feeds never leave your local area network. No video frames, bounding boxes, or personal data are ever uploaded to cloud servers or third-party tracking services.
Polygon Zones & Tripwire Rules
Define custom polygon detection zones, perimeter lines, and exclusion masks directly on camera preview feeds to eliminate false alarms and target specific areas of interest.
Instant Low-Latency Local Alarms
Trigger immediate Android native alarms, ringers, and persistent notification alerts the instant a zone rule is broken, with instant high-resolution frame snapshots stored locally.
On-Device Benchmark & Model Hub
Download, validate, test, and benchmark local models side-by-side. Inspect real-time frame rates, memory consumption, and inference milliseconds directly on your target hardware.
Triple Native Inference Engines
GYN lite bypasses high-overhead interpreted runtimes. Inference and post-processing execute entirely inside the Android native C++/Kotlin layer across three first-class backends via our unified engine contract.
Google LiteRT Engine
TFLite Successor • GPU Shader Pipeline
Broad compatibility across modern Snapdragon, Dimensity, and Exynos chipsets.
Tencent NCNN Engine
High-Performance Mobile C++ & Vulkan
Maximum FPS throughput and lower battery thermals under intensive multi-stream loads.
Microsoft ONNX Runtime
Enterprise NNAPI & Direct GPU Acceleration
Next-gen architectures like YOLOv10 with end-to-end NMS-free direct inference.
End-to-End Edge Pipeline Architecture
RTSP, IP Cam, NVR, DVR, or CameraX sensor
Letterbox scaling, NV21/YUV to RGB tensors
LiteRT GPU, NCNN Vulkan, or ONNX NNAPI
NMS filtering, polygon entry & tripwire checks
Foreground audio siren & forensic snapshot
Connect Any Camera, NVR, or DVR
No proprietary hubs or vendor-locked camera ecosystems. If your system can output an RTSP or local video stream, GYN lite can run real-time computer vision analytics on it.
RTSP Live Video Feeds
Direct connection to standalone IP cameras, streaming encoders, or software feeds via standard RTSP URIs with credentials and transport overrides.
NVR Recorders (Network Video Recorders)
Ingest discrete channel streams from Hikvision, Dahua, Reolink, UniFi, or generic NVR backends directly into dedicated GYN lite monitoring jobs.
DVR Units (Digital Video Recorders)
Modernize legacy analog/coaxial DVR systems with edge computer vision AI by streaming channel RTSP endpoints through local IP network bridges.
On-Device Mobile Cameras
Turn spare Android phones and tablets into standalone smart security sentinels using the device's high-resolution front or rear cameras.
Interface Walkthrough & Features
Engineered with a streamlined Material 3 dark design system, intuitive job wizards, and uncompromising speed.
Centralized Monitoring Command
Instant overview of running foreground jobs, active camera feeds, inference FPS rates, and detection events with quick start/stop orchestration.
Dashboard & Live Monitor UI
Reserved display slot ready for official application screenshots.
Authentic Brand Badge & Wordmark
The canonical G[logo]N lite brand system engineered with interactive optical reticle geometry.
Edge Vision Intelligence
100% Offline • On-Device Neural Processing • Zero Cloud
Dummy UI / Computer Vision Sandbox
Test detection threshold filters, toggle inference backends (LiteRT, NCNN, ONNX), and observe how bounding boxes and tripwire calculations respond in real-time.
Video Guides & Documentation
Comprehensive tutorials, setup walkthroughs, and deployment blueprints.
GYN lite: Complete 5-Minute Setup & RTSP Deployment
Video production in progress. This slot is reserved for high-definition video embeds (YouTube / PeerTube / self-hosted video players).
Connecting an IP Camera via RTSP
Step-by-step walkthrough of finding your camera's RTSP endpoint and configuring sub-streams.
Optimizing LiteRT vs NCNN for Battery Life
How to tune frame interval skips, INT8 quantization, and GPU delegates for 24/7 monitoring.
Drawing Custom Polygon Tripwires
Configuring boundary detection rules to avoid false positives from sidewalks or trees.
Connect With The GYN Ecosystem
Join the developer network building autonomous, local-first computer vision surveillance.
GitHub Repository
Open Source Code
Inspect native C++/Kotlin inference engines, submit pull requests, and file issues.
Discord Community
Real-time Chat
Chat with developers, discuss camera setups, model quantization, and edge hardware tweaks.
X / Twitter
Announcements & Releases
Follow real-time development updates, benchmarks, and new engine feature announcements.
Hugging Face Models
Model Weights & Quantization
Download pre-quantized LiteRT and NCNN computer vision models optimized for edge devices.
Frequently Asked Questions
Answers to common questions about on-device inference, compatibility, and privacy.
Ready for Autonomous On-Device
Computer Vision Monitoring?
Transform any Android device into an intelligent edge sentinel. Zero cloud accounts, zero recurring fees, and absolute privacy for your RTSP feeds, NVRs, and DVRs.