Hailo-15 is a family of high-performance edge AI vision processors designed for smart cameras. The Hailo-15 System-on-a-Chip (SoC) combines Hailo’s world leading AI inferencing technology with advanced computer vision engines and AI-powered ISP to deliver premium image quality and sophisticated video analytics.
Purpose-built for on-device inference, Hailo-15 enables the efficient execution of multiple complex deep learning applications alongside AI-powered image enhancement. While model training remains in the cloud, inference is performed locally with exceptional efficiency, delivering high AI compute performance per watt and per dollar. This allows smart cameras to comply with strict size, cost, and power constraints without compromising on scale or capability.
Superior Image Quality
Classic + AI-ISP for video enhancement in extreme low light conditions (<0.01LUX)
Advanced Analytics
High compute power for complex pipelines of video analytics and generative AI
AI-Centric
World’s leading edge AI architecture embedded in vision SoC
Efficiency
Easy-to-use, open SDK
Embraced by the world’s largest edge AI community
Enhance dark, noisy scenes in real time at the edge. Our low-light pipeline denoises, boosts exposure, and preserves colors and details while minimizing motion blur, in extreme low light conditions <0.01 Lux. Ideal for night shots, indoor surveillance, and dim industrial environments providing crisp and usable video without increasing power consumption, latency, or bandwidth.
Market leading HDR – capture details and accurate colors even at challenging light conditions in scenes with extreme bright and dark areas. Hailo’s advanced computer vision engines and AI-ISP handles merging both long and short exposures to deliver crisp and clear images achieving high dynamic range with minimal motion blur
Unlock real-time image understanding that matches live video frames to natural-language prompts. No task-specific training required. The CLIP Zero-Shot Classification app brings flexible, promotable recognition to your edge projects, ideal when speed and adaptability matter.
Protect identities in real time, right on the edge. DPM automatically detects and masks sensitive regions in live video, with configurable zones and blur/blackout styles. Maintain analytics quality while meeting privacy requirements, all with ultra-low latency and no cloud dependency
The Hailo-15 series of AI Vision Processors is designed to meet the requirements of a diverse range of cameras, by offering 2 products:
The Hailo-15H, the high-performance member of the Hailo VPU family, is optimized for mid-to-high-end intelligent surveillance IP cameras
Delivers unprecedented image quality and advanced AI analytics for mass market smart cameras, offering an exceptional balance of image quality, AI analytics performance, and cost efficiency
Up to 20 TOPS (Tera Operations Per Second) running on a powerful Neural Network (NN) Core enabling processing of multiple advanced DL models in parallel
High FPS DL model processing for faster and highly accurate detection of more objects per frame
Best AI performance at a standard camera power consumption and cost envelope
AI-powered vision processing with low–light, image stabilization and image perfection capabilities, for enhanced image quality
Premium 4k30 image quality with a cutting-edge ISP pipeline and an advanced Vision sub-system with High Dynamic Range (HDR) and Noise Reduction (NR) algorithms
Industry leading DSP vision processing and high-quality video encoding
Industry-standard infrastructures and frameworks with a complete Yocto-based Linux distribution runtime on the Application Processor
Industrial grade SoC with a variety of interfaces for image sensors, data, and memory
Secure boot and debug with hardware accelerated crypto library, TrustZone, TRNG, and Firewall
A complete set of VPU drivers, libraries, and tools to accelerate development of AI-powered smart cameras
Full support for camera hardware interfaces and protocols, combined with specialized image and video processing algorithms
Optimized for camera-based workloads to deliver efficient execution, seamless system integration, and advanced computer vision functionality
