







Horus AI Camera
• AI camera modules based on Rockchip RV1126
• Quad core ARM Cortex A7 processor
• Support NNIE performance up to 2.0Tops
• Offer USB2.0 x 1/Ethernet port x 1/UART x 2/TF card expansion interface
• Accelerates AI algorithm hardware deployment
• IPC 38-Board standard structure design, quick on-demand customization of enclosures
The Horus AI Camera provides a highly customizable software and hardware development platform, featuring a Rockchip RV1126 processor solution, optional Sony IMX series sensors, POE power supply, and data transmission. Horus adopts an IPC 38-Board standard structure design, with a shell that can be quickly customized on demand and support for the entire development cycle covering development, verification, and mass production. Horus includes a complete SDK, equipped with system, driver, image processing, and AI application interfaces. The platform supports extensive development and pre-research in Python, C, and C++, accelerating the implementation cycle from AI algorithm development to hardware deployment.
System Block Diagram
Application Scenarios
The Horus AI Camera is extensively used in applications requiring advanced visual recognition, such as facial recognition, body shape recognition, passenger flow analysis, license plate recognition, fatigue testing, helmet recognition, and OCR recognition.
Facial Feature Recognition
Horus excels in multi-dimensional facial feature recognition, making it ideal for monitoring on-duty personnel and workers. It can analyze various behaviors and states, such as fatigue level, activity level, smoking, yawning, and phone usage. The system not only comprehensively evaluate the working status but also has the capability to upload data to a data center or trigger local alarms to prompt immediate adjustments in working status.
Human Body Feature Recognition
Horus excels at multi-lens and multi-position tracking in crowded environments such as specialty stores, exhibition halls, and business centers. The system captures and analyzes faces, comparing them with face databases in the system to distinguish between returned and new customers. Additionally, it recognizes other distinctive human features, such as clothing and accessories, using cross-regional markings to enable comprehensive tracking and analysis of customer behavior, thereby improving store layout and operational efficiency.
Hard Hat Recognition
Leveraging large-scale training on safety helmet image datasets, Horus is proficient at automatically recognizing the proper wear and standardization and wearing status of helmets on workers, enabling efficient and effective safety supervision, promoting safety practices on construction sites and in other work environments.
Specifications
Operating System |
Linux |
CPU |
Quad core ARM Cortex A7 @ 1.5 GHz |
RAM |
1 GB DDR3 SDRAM |
Memory |
8 GB EMMC, support TF card storage expansion |
NPU |
2.0 Tops, support INT8/INT16 |
Graphics and Image Processing Capabilities |
Maximum resolution: 14 M (4416 X 3312) Support ISP 3A function: AE, AF, AW Support HDR high dynamic range rendering, tone mapping Support 3D, 2D noise reduction, image enhancement, sharpening Support lens distortion and fisheye correction, Gamma correction Support feature point detection |
Video Codec |
1. H.264/H.265 maximum resolution of codec 4096 x 2304 @ 30 fps 2. H.264/H.265 multi-stream real-time encoding (decoding) capability, -3840 x 2160 @ 30 fps coding + 1080p @ 30 fps coding -3840 x 2160 @ 30 fps coding + 3840 x 2160 @ 30 fps coding |
Watchdog |
CPU built-in watchdog |
Sensor Interface |
Support photosensitive sensor interface |
Firmware Encryption |
Support firmware encryption |
RTC |
External |
Board Size |
42 x 42 (MM) (IPC 38-Board screw hole distance) |
Operating Temperature |
-10°C to 50°C |
Single-Board Working Voltage |
9 - 16 V, typical value 12 V |
Lightning Protection |
Network port supports 4 KV lightning protection |