MistySOM is a System on Module that provides High Performance with Low Power
MistySOM is a battery-powered system-on-a-module manufactured by MistyWest, a company known for its engineering products and consultancy services. It is developed to be used for AI-based applications that specifically require high processing capabilities and low power consumption. In the traditional SOMs, this was a tricky combination to achieve, especially with battery-based designs. SOMs were unable to provide low-power applications. Also, in the systems with higher power consumption there was overheating which meant there would be a bulky heat sink. The new MistySOM consumes half the power as the other microprocessors, for the same computer vision tasks. It is built to enable battery-powered computer vision applications and solve the challenging power problems that are common in today’s IoT environments.
It uses the power of Renesas’s series of RZ/V2L processors to meet the demands of modern IoT edge devices, especially for image processing applications. It fits perfectly with their highly capable and energy-efficient AI chips. The Renesas RZ/V2L is a dual-core Arm Cortex-A55 CPU that runs at 1.2 GHz. It also features the Dynamically Reconfigurable Processor (DRP) Technology that allows a vision algorithm to change configurations on the go. MistySOM can provide a reduction in charge cycles and a reduced bandwidth requirement. The DRP-AI exhibits excellent power efficiency and eliminates the need for any additional heat dissipation measures, such as heat sinks or cooling fans, to achieve consistent performance.
Specifications of MistySOM
- Low power requirements
- Yocto Linux Operating system
- AI accelerator
- Cortex-A55 (Dual or Single)
- 3D graphics engine (Arm Mali-G31)
- Video codec (H.264)
- Camera interface (MIPI-CSI)
- Display interface (MIPI-DSI)
- Two USB2.0 interfaces
- Two CAN interfaces
- Gigabit Ethernet 2ch
- 2GB DDR4 RAM
- Supports WiFi and Bluetooth interfaces on the I/O board
- It is compatible with the Renesas RZ/G2L processor for less demanding applications
Using the RZ/V2L, MistySOM can perform a number of AI computer vision processing such as graph optimization and FP16 quantization. It has the capabilities of performing these processes at high speeds and with a minimal power budget, with the help of its in-built AI accelerator DRP-AI. Moreover, with a dedicated Neural Processing Unit, MistySOM can perform AI computer vision tasks at 50% less power than the other processors available. It delivers high-speed AI inference at low power consumption, backed by the NPU that supports standard ONNX ML models.
A trained ONNX model can be implemented cost-efficiently, enabling customers to implement a variety of AI-based vision applications without requiring an external image signal processor (ISP). The images can be captured through the MIPI-CSI interface and h.264 encoded. This integrated combination allows MistySOM to provide a real-time AI inference engine with hardware-accelerated image processing functions, like color correction and noise reduction. It is ideal for any application that requires object detection and image classification in a small battery-powered system.
To learn more about it, visit the product page