Hailo launches a new vision processor unit for smart cameras at the edge
The new Hailo-15 processor can deliver up to 20 TOPS. Source: Hailo
Hailo, a company that makes chips for artificial intelligence, has introduced Hailo-15, a new vision processor unit for smart cameras. The vision processor has various applications, such as manufacturing, Smart Cities and retail locations, and provides high-performance video processing and analytics at the edge. The onboard system-on-chip uses advanced deep learning models that enable AI inference with a neural network core, delivering a performance of up to 20 tera operations per second running.
Hailo also unveiled a camera development kit that uses the Hailo-15 processor. According to the company, this kit includes all the necessary tools for prototyping machine vision applications. The development kit consists of a single-board computer that supports a dual sensor interface, data, communication and display interfaces. On the software side, the platform features a Yocto-based Linux distribution, a full stack DSP, libraries and drivers.
“Hailo-15 represents a significant step forward in making AI at the edge more scalable and affordable,” says Orr Danon, CEO of Hailo. “With this launch, we are leveraging our leadership in edge solutions, which are already deployed by hundreds of customers worldwide; the maturity of our AI technology; and our comprehensive software suite to enable high-performance AI in a camera form factor.”
Recommended reading: Hailo raises $136 million to further development of AI-powered edge solutions
Hailo claims the Hailo-15 vision processor chip uses AI techniques and image stabilization features to provide superior image quality, even in low-light conditions. This chipset can handle multiple input streams at 4K with 60 frames per second image quality, incorporating an ISP pipeline and a vision sub-system with high dynamic range and noise reduction algorithms. The Hailo-15 vision processor chipset is available in three versions — Hailo 15H, Hailo 15M and Hailo 15L, which offer 20 TOPS, 11 TOPS and 7 TOPS performance, respectively.
According to Hailo, the Hailo-15H chipset, which has the highest performance metrics among the three versions, can process the YoloV5M6 object detection model with an input resolution of 1280×1280 pixels at a real-time sensor rate. Additionally, the company states the chip can run the ResNet-50 model at 700 frames per second.
“By creating vision processors that offer high performance and low power consumption directly in cameras, Hailo has pushed the limits of AI processing at the edge,” says KS Park, head of R&D for Truen, specialists in edge AI and video platforms. “Truen welcomes the Hailo-15 family of vision processors, embraces their potential, and plans to incorporate the Hailo-15 in the future generation of Truen smart cameras.”
Hailo will demonstrate the Hailo-15 vision processor unit at the ISC-West in Las Vegas, Nevada.
Article Topics
AI/ML | computer vision | edge AI | Hailo | processor | retail | smart cameras | smart cities | Truen
Comments