Smart Recognition Solution

Our smart recognition solution supports IPC smart recognition algorithms, face recognition algorithms and palm vein recognition algorithms, enabling object-of-interest snapshot capture, license plate recognition, face recognition and palm vein recognition. It can be used in all kinds of IPC cameras and access control terminals, and offers a high capture rate, a low false-capture rate, high recognition accuracy and fast recognition speed.

Recommended Devices

• RV11 series: RV1126: quad-core Arm® Cortex®-A7 plus RISC-V MCU core architecture; 150ms fast boot; 2.0Tops NPU, support INT8/ INT16; 14M ISP with 3-frame HDR; supports 3 cameras input simultaneously; 4K H.264/H.265 video encoding and decoding

Key Features

• Developed with the open-source GUI framework AWTK, with a mature demo available
• Complete development documentation and full technical support
• One hardware platform: by fitting different cameras and loading different software packages, it can perform IPC smart recognition, face recognition or palm vein recognition

Applications

Smart locks, IP cameras, dash cams, interactive gaming

Reference Design IPC Smart Recognition Demo Kit

• Face & body function: overall face capture rate 99.5%, overall human body capture rate 97.5%
• Face & body function: up to 128 faces captured on one screen, up to 5 recognized on one screen
• Structured analysis function, motion speed of detected objects:
• • Motor vehicle / license plate: <50km/h, plate visible for more than 1s
• • Face / human figure: <5km/h
• • Non-motor vehicle: <20km/h

Reference Design Face Recognition Demo Kit

• Supports face detection and tracking
• Supports feature extraction and 1:1 / 1:N matching
• Supports attribute recognition such as safety helmet, age and gender
• Supports a 100,000-record database, with the full process taking <550ms on a 100,000-record database
• Supports face recognition with a face mask worn
• Full process <150ms on a 10,000-record database
• Supports software / hardware licensing

Reference Design Palm Vein Recognition Demo Kit

• Contactless recognition: captures the user's palm vein image to extract vein features for live-body recognition
• Requires the user to actively present the palm; data is encrypted, giving a high security level
• Feature extraction time: 25ms
• Feature matching time: 4ms
• Recognition accuracy: FRR<0.001%, FAR<0.00001%
• Recognition distance: 10-20cm