EmbeddedRelated.com
The 2026 Embedded Online Conference
Kria KV260 Vision AI Starter Kit

Kria KV260 Vision AI Starter Kit

MPN: SK-KV260-G
AMD (Xilinx)
FPGA-based edge AI starter kit useful for hardware-accelerated vision and ML prototyping.
Active450 in stock

Overview

The Kria KV260 is an out-of-the-box development platform for the K26 System-on-Module (SOM), specifically designed for high-performance vision AI applications. It combines a Zynq UltraScale+ MPSoC with dedicated interfaces for multi-camera inputs and video outputs, enabling hardware-accelerated machine learning at the edge. The kit simplifies the transition from prototype to production by using the same SOM hardware found in industrial deployments.

Why Choose This Part

The platform offers a significant performance-per-watt advantage for vision tasks by offloading compute-heavy ML layers to the FPGA fabric. It includes pre-built accelerated applications and supports common AI frameworks like TensorFlow and PyTorch through the Vitis AI environment. Hardware security is prioritized with a dedicated TPM 2.0 module and Secure Boot capabilities for trusted edge deployments.

Applications

Smart City Surveillance
Utilizing the multi-camera MIPI interfaces and onboard ISP for real-time traffic monitoring and pedestrian counting.
Industrial Machine Vision
Implementing high-speed defect detection and automated optical inspection on assembly lines using FPGA-based acceleration.
Retail Analytics
Running heat-mapping and customer behavior models locally via the edge AI optimized Processing System.
Robotic Navigation
Processing spatial data and SLAM algorithms using the integrated Zynq architecture for low-latency movement control.

Key Specifications

Type FPGA + MCU/MPU SoC
Contents Board(s)
Platform Kria KV260 Vision AI Starter
Utilized IC / Part SM-K26, XCK26

Getting Started

Start by downloading the Kria SOM starter disk image and flashing it to a microSD card. Utilize the Xilinx Vitis AI development environment to deploy models to the DPU (Deep Learning Processor Unit) inside the FPGA fabric. The board connects to a host PC via the JTAG/UART port for debugging and supports 12V power input for operation.

The 2026 Embedded Online Conference