
Intelligent Vision Products
Combining core algorithms such as AI Super-Resolution, AI Video Stabilization, AI Color Enhancement, and AI ISP tuning, Attrsense Intelligent Vision Products deliver full-stack vision capabilities, from edge SoCs to SDK toolkits, so devices keep producing clear, stable images under challenging lighting, motion, and low-bandwidth conditions.
INTELLIGENT VISION SOC

Super-Resolution ChipATTRSENSE VISION-SR SOC
The AI SR SoC pairs a high-performance NPU with deep super-resolution algorithms. Its on-device, real-time super-resolution hardware upscales 2MP → 4MP in real time for ultra-HD image quality. Built-in spatial information fusion and temporal motion compensation suppress upscaling artifacts, color banding, and other noise while preserving detail in dynamic scenes, keeping images clear and stable even in extreme low light and high-speed motion.

Imaging ChipATTRSENSE VISION-ISP SOC
The AI ISP SoC combines an efficient NPU with a tightly coupled ISP pipeline. Deep learning–based, in-house denoising and color enhancement modules significantly improve dynamic range, resolution, signal-to-noise ratio, and color in low-light scenes. Global and local adaptation lets it output the best image for every scene, with high-fidelity results in night vision, backlit, and high-contrast conditions.
RECOMMENDED CHIPS
Specifications
- CPU
- Master - up to 900 MHzSlave - up to 450 MHz
- NN
- 0.5 TOPS
- ISP
- 2DNR, 3DNR, Y/CNR, DRC, AE/AWB
- Encoder/Decoder
- H.264 / H.265 video encoder- 2880×1620@30fps H.264/H.265- 2560×1920@25fps H.264/H.265JPEG encoder- 2880×1620@30fps- 2560×1920@25fps
- Ethernet
- Internal 10M/100M FE PHY
- Memory
- Internal 64MB DDR2: up to 1333 Mbps
- Package
- QFN, 9mm × 9mm
Key Features
• Detect & Search Anything
- -- Vector search with real-time results (4K@15fps)
- -- Fuzzy semantic search with long-context understanding
• AI SR
- -- 2M sensor matches 4M sensor clarity
- -- Generalizes to faces, license plates, pets, and more
• AI ISP
- -- Starlight-grade denoising, full-color night vision at 0.001 Lux
Main Modules
- Master CPU
- Slave CPU
- ISP
- Decoder
- Encoder
- 64MB DDR2
- 0.5 TOPS
- USB2.0
- FE PHY
INTELLIGENT VISION SDK

Super-Resolution SDKATTRSENSE VISION-SR SDK
The SR SDK combines an efficient super-resolution engine with image enhancement modules and supports 2×/4×/8× upscaling. Where bandwidth is limited or storage budgets are tight, it adaptively enhances image quality in real time, intelligently improving texture, clarity, color, and sharpness to suit different viewing needs, with no visible degradation across lighting conditions or scene complexity.

Stabilization SDKATTRSENSE VISION-EIS SDK
Built on AI motion estimation and intelligent compensation algorithms, the EIS SDK delivers powerful video stabilization without any added mechanical parts. Its multimodal dynamic scene perception model is compatible with a wide range of devices and is used in action cameras, smart hardware, drones, and more, keeping footage stable and smooth even under complex shake.
PRODUCT ADVANTAGES
Integrated, Fast to Deploy
Standardized SDK interfaces support multiple platforms (Android / Linux / Windows) and embed quickly into existing products without changes to the original architecture.
Built for Many Scenarios
Works across cameras, mobile devices, cloud platforms, edge devices, and more, flexibly serving security surveillance, AI vision, smart hardware, and cloud image processing.
Hardware-Software Synergy
Deep integration of AI ISP, NPU, and multimodal algorithms enables ultra-HD reconstruction and stabilization on the device itself, improving image quality and stability together.
APPLICATION SCENARIOS

Motion Stabilization
Powered by: AI Super-Resolution + Motion Estimation + AI Color Enhancement
Scenario Value
Fixes shaky, blurry, and cold-toned footage from moving cameras, using intelligent stabilization to output clear images.

Clear Night Imaging
Powered by: AI ISP + AI Super-Resolution + AI Color Enhancement
Scenario Value
Improves visual quality in noisy, backlit, and low-light scenes, with colors closer to what the eye sees.

HD over Weak Networks
Powered by: AI Super-Resolution + Adaptive Encoding
Scenario Value
For video over 4G and CAT.1 networks, AI super-resolution inference restores 4K HD quality while optimizing bandwidth, reducing transmission cost and latency.