The Hailo-15L AI Vision Processor HNC2LAI11SM is a cost-efficient edge AI System-on-Chip developed specifically for mass-market and high-volume smart camera applications. It combines dedicated AI inference hardware, an advanced image signal processor, application processing, DSP vision processing and camera interfaces within a single integrated device.
The Hailo-15L provides up to 7 TOPS of neural-network inference performance, enabling multiple advanced deep-learning models to operate directly on the camera. Local AI processing allows smart-camera systems to perform object detection, classification, video analytics and other computer vision workloads without continuously sending raw video to cloud infrastructure.
Hailo designed the Hailo-15 family around an integrated AI-powered ISP and advanced vision subsystem. The Hailo-15L supports high-quality 4K30 image processing, including low-light enhancement, high dynamic range processing, image stabilisation and advanced noise reduction. Hailo specifically positions the processor for high-quality imaging in extremely low-light environments below 0.01 lux.
The image-processing architecture supports HDR, advanced 2D and 3D noise reduction, chroma noise reduction, lens shading correction, distortion correction, digital zoom, flip, rotation and electronic/digital image stabilisation. This allows camera manufacturers to combine conventional image processing with AI-based image enhancement within the same SoC.
An integrated Vector DSP provides additional vision-processing capability, while the application-processing subsystem uses a quad-core ARM Cortex-A53 processor. For the Hailo-15L configuration, Hailo specifies the application processor at up to 1.1 GHz, providing approximately 10.1 kDMIPS of processing performance.
The Hailo-15L supports LPDDR4 memory, with the Hailo-15L configuration specified at up to 3200 MT/s. Hailo also highlights minimal DRAM requirements, with supported smart-camera implementations using as little as 1GB of DRAM, helping manufacturers reduce system cost and board complexity.
For camera connectivity, the processor provides dual MIPI CSI video inputs and a 24-bit DVP interface, along with MIPI DSI video output. Additional peripheral interfaces include I2C, I2S and SPI for connection to sensors and supporting devices.
The Hailo-15L also provides PCIe Gen 3 connectivity, Gigabit Ethernet, USB and storage interfaces, enabling integration into complete IP-camera and edge-vision architectures. For the Hailo-15L configuration, the PCIe interface supports up to two Gen 3 lanes.
Integrated video encoding supports H.265/HEVC and H.264/AVC, including multiple video streams. This makes the processor suitable for network cameras, security systems, smart-city infrastructure and other applications that require simultaneous AI analytics and compressed video streaming.
Security features include secure boot, secure debug, TrustZone, TRNG, firewall protection and a hardware-accelerated cryptographic library, helping camera manufacturers develop secure connected vision products.
The Hailo-15L operates with power consumption below 3 W, helping it meet the thermal and energy requirements of compact camera designs. The industrial device supports operation from -40°C to +85°C and uses a compact 13 × 13 mm package.
Hailo provides a comprehensive Vision Processor Software Package including a Yocto-based Linux environment, drivers, media libraries, camera applications, HailoRT and the Hailo Dataflow Compiler. This integrated software environment helps reduce development effort for AI-enabled smart cameras and embedded vision products.
