JP1126 Binocular Face Recognition 3D Stereo Vision Camera Module with Algorithm

JP1126 Intelligent Dual-Lens Camera Module with Algorithm

Product Overview

The JP1126 Intelligent Dual-Lens Camera Module with Algorithm is a complete turnkey facial recognition solution built around the RV1126 AI vision processor — a quad-core 32-bit ARM Cortex-A7 architecture running at 1.5GHz, with an integrated NPU delivering up to 2.0 TOPS of AI computing power. This all-in-one module combines a 2MP RGB + IR dual-lens camera with onboard face recognition algorithms, supporting databases of up to 100,000 users with >99.7% recognition accuracy and fast liveness detection (15ms for binocular) to effectively prevent photo, video, and wax mask spoofing. The module handles all processing locally, eliminating the need for external host computing. With 4K video encode/decode capabilities, RKNN model conversion tools supporting Caffe, Darknet, MXNet, ONNX, PyTorch, TensorFlow, and TFLite, and compact aluminum-alloy housing with integrated heat sink, the JP1126 enables rapid, low-threshold deployment of facial recognition terminals for access control, smart finance, construction site management, and smart mobility applications.

Key Advantages

• 2.0 TOPS NPU for On-Device AI: Built-in neural processor delivers 2.0 TOPS of INT8 performance — enabling fast, accurate face detection (~23ms), tracking (~7ms), feature extraction (~25ms), and comparison (~0.0115ms per match) entirely on-device without cloud dependency.

• Complete Turnkey Solution: Fully integrated module with onboard RV1126 processor, dual-lens camera, pre-loaded recognition algorithms, and SDK — no external host, no complex integration. Connect power and USB, and start deploying facial recognition immediately.

• 100,000-User Capacity: Supports up to 100,000 face templates with recommended database of 10,000 — scalable from small office access control to enterprise-level deployments, with 99% recognition accuracy at 0.01% FAR under standard testing conditions.

• Binocular Liveness Detection in Just 15ms: RGB + IR dual-camera system with binocular liveness detection completes spoof checks in approximately 15ms — effectively blocking photos, videos, 3D masks, and wax models without compromising user experience.

• 4K Video Encode/Decode: Supports 4K H.264/H.265 encoding at 3840×2160@30fps with simultaneous 720p@30fps encoding — enabling high-resolution video recording and streaming alongside recognition processing.

• Broad AI Framework Compatibility: RKNN model conversion tools support Caffe, Darknet, MXNet, ONNX, PyTorch, TensorFlow, and TFLite — enabling easy migration of custom AI models and algorithm integration.

• Compact, Thermally Efficient Enclosure: Aluminum alloy body with serrated heat sink back cover measuring just 84.0mm × 22.45mm × 19.35mm — dissipates heat efficiently while maintaining a small footprint for space-constrained terminal designs.

• Flexible System Support: Runs on Linux with support for Android 7–10 — enabling deployment across a wide range of hardware platforms and operating environments.

Why Choose This Module?

• All-in-One: No Host, No Cloud, No Complexity: The JP1126 is not just a camera — it's a complete facial recognition computer. With onboard RV1126 processor handling all detection, tracking, liveness checking, feature extraction, and comparison, there's no need for external hosts, cloud services, or additional processing hardware. Connect it to a display, and you have a working face recognition terminal. This dramatically reduces development time, BOM cost, and system complexity.

• 10,000–100,000 Users Without Breaking a Sweat: Many modules claim "large database support" but slow down significantly beyond a few hundred users. The JP1126 maintains consistent sub-25ms feature extraction and sub-0.012ms comparison speeds across its entire 100,000-user capacity. This means no user-facing performance degradation as your deployment scales — critical for enterprise installations with large user populations.

• 15ms Liveness Detection — The Speed Matters: While binocular liveness detection typically takes 45ms or more, the JP1126 achieves ~15ms — over three times faster. In high-traffic environments where hundreds of users pass through daily, this speed difference is the gap between seamless flow and frustrating bottlenecks. Users won't notice the liveness check; they'll only experience fast, reliable access.

• Wear a Mask? No Problem: The JP1126 maintains 95% recognition accuracy with masks — essential for post-pandemic access control environments where mask-wearing remains common. With a 10,000-person database and 0.01% FAR, the module delivers reliable recognition whether users are masked or unmasked.

• Built to Handle Real-World Conditions: With support for pitch/yaw/roll angles up to ±30°, minimum face size of 50×50 pixels (90×90 with liveness), and a -10°C to 60°C operating range, the JP1126 accommodates real-world usage variations — users of different heights, at different angles, and in varying environmental conditions. This reduces friction and improves throughput in busy access control points.

HD Face Recognition Module

Face Recognition Module DC5V

1920x1080 face detection module

JP1126 Binocular Face Recognition 3D Stereo Vision Camera Module with Algorithm Images