May 27, 2025
Article
We are excited to highlight the successful integration of the Vitis AI NPU on the Versal VE2302 System on Module and Evaluation Kit. Leveraging the power of the iW-RainboW-G57D System on Module (SoM), this initiative underscores iWave’s commitment to accelerating innovation in EdgeAI. The implementation provides a high-performance, low-latency object detection model on an FPGA-powered edge AI system that can efficiently process real-time data streams while maintaining power efficiency and computational speed.
The Vitis is an AI inference acceleration software designed to run AI models efficiently on FPGA-based hardware, particularly AMD FPGAs and Versal Adaptive SoCs. It enables GPU-to-FPGA transition, allowing AI models trained for GPUs to run directly on FPGAs without modification. Supporting frameworks like TensorFlow, PyTorch, ONNX, and Caffe, it ensures seamless AI deployment.
Explore more about the demo of integration of the NPU Stack on the Versal AI Edge.
The AI inference workflow on iW-RainboW-G57D, leverages the VART X API modules and structured for real-time processing,. A script is created to run the NPU application on the SoM. This script captures video using the USB camera and converts it to NV12 format. The converted video is then processed by NPU and detected objects are highlighted on the HDMI display.
Dataflow process is as follows:
Experience real-time object detection on VE2302 Versal AI Edge SoM! Watch the demo as iWave experts showcase AI inference using NPU IP for precise and efficient object detection.
The Versal AI Edge provides a single heterogeneous platform leveraging programmable logic for sensor fusion, AI Engines for AI inference compute, and a processing system for real-time control. Both AI Engines and DSP Engines for AI, vision processing and radar & LiDAR processing. The DSP Engines for diverse workloads including image signal processing, support for single-and half-precision floating point. The Versal Adaptive SoC also supports Native MIPI support to handle up to 8- megapixel resolutions and beyond.
The Versal AI Edge-based System on Module (SoM), compatible with VE2302, VE2202, VE2102, and VE2002 chips, is designed for high-performance edge AI applications. It features 8GB LPDDR4 RAM, 128GB eMMC, and 256MB QSPI Flash, along with two high-speed expansion connectors and 122 user-configurable I/Os for seamless integration. With 28.21Gbps transceiver blocks, 40G multi-rate Ethernet, PCIe, and native MIPI support, the SoM ensures high-speed connectivity and efficient data processing, making it ideal for vision-based AI and other advanced edge applications.
iW-RainboW-G57M solution revolutionizes edge AI computing by delivering high-performance inference, low latency, and scalable connectivity. With its powerful AI acceleration, real-time processing, and robust security, the SoM is an ideal choice for AI-driven applications. iWave provides a robust suite of tools, libraries, and software resources that empower developers to harness the full potential of the Versal AI Edge System on Modules.
Contact us at mktg@iwave-global.com to explore how the Versal AI Edge SoM can revolutionize your AI solutions!
For more information visit www.iwave-global.com.
We appreciate you contacting iWave.
Our representative will get in touch with you soon!