Hailo, an Israeli startup focused on edge artificial intelligence (AI) processors, today launched a new System-on-a-Chip (SoC) for smart cameras that promises industry-leading AI analytics in a camera power envelope. The Hailo-15 family of vision processors are designed for direct integration into smart cameras in a wide range of applications, from industrial environments to retail locations.
Enabled by a powerful Neural Network (NN) core that can deliver up to 20 tera operations per second (TOPS), the Hailo-15 is able to simultaneously and efficiently process multiple complex deep learning (DL) applications at full scale. This increased AI capacity allows for improved video quality in low light, video enhancement and stabilization, and high dynamic range performance. Plus, this SoC can support multiple input streams at 4K resolution.
Hailo-15’s AI achieves fast object detection at a high resolution, so that it can identify smaller and more distant objects with increased accuracy and fewer false alarms. The chip uses vision-based transformers in the camera for real-time object detection. The Hailo-15 can run the object detection model YoloV5M6 with an input resolution of 1280×1280 at real-time sensor rate, or ResNet-50, the industry classification model benchmark, at 700 frames per second (FPS).
The Hailo-15 comes in three variants—the 15H, 15M, and 15L—to address various processing needs and price points. The models range from 7 TOPS to the aforementioned remarkable 20 TOPS, delivering over five times higher performance than other, comparably priced solutions on the market. This is perhaps especially enticing to customers as Hailo-15-integrated cameras can lower the costs of large-scale camera deployments by offloading cloud analytics to save video bandwidth and processing.
Notably, the Hailo-15 also consumes little power and is hence suitable for any type of IP camera. This also means manufacturers can design fanless edge devices around the Hailo-15, ideal for industrial and outdoor solutions.
Sam Wallace - became Circuit Cellar's Editor-In-Chief in August 2022.
His experience in writing, editing, and teaching will provide a great perspective on the selection, presentation, and clarity of editorial content. The Circuit Cellar audience will benefit from his strong academic background encompassing a Master of Fine Arts in Writing and a Bachelor of Science in Mathematics with honors. His passion for learning and teaching is a great fit for Circuit Cellar's continuing mission of Inspiring the Evolution of Embedded Design.