The Edge Vision SoC (EVSoC) is a complete, ready-to-deploy embedded framework for building next-generation vision, imaging, and display systems on Efinix® FPGAs. Built on the Sapphire RISC-V SoC, the EVSoC framework seamlessly integrates camera capture, ISP processing, memory management, compute, and real-time display output—reducing development complexity and accelerating time-to-market.

Edge Vision SoC Framework

A Complete Vision Framework

Edge AI and embedded vision are enabling a new generation of intelligent systems, delivering real-time decision-making across applications ranging from automotive and industrial automation to retail and robotics. To accelerate development, the Efinix EVSoC framework provides ready-to-use software drivers, and fully validated example designs for building complete vision systems.

Features:

  • Camera Subsystem

    • Complete camera pipeline with sensor capture, integrated ISP functions (colour gain, demosaic, sobel, dilation, erosion), and real-time frame processing.

  • Display Subsystem

    • Real-time display output with flexible timing control and efficient frame handling for responsive and high-quality edge vision applications.

  • Memory Subsystem

    • High-throughput memory buffering and efficient frame data transfer to support real-time vision processing and smooth system performance.

  • Flexible AXI4/APB3 System Interconnect

    • Enables modular connectivity between processing, memory, peripherals, and custom IP for scalable and extensible edge vision designs.

  • Sapphire RISC-V Control and Software Framework

    • Provides software-driven control of sensor streaming, frame handling, image enhancement, ISP tuning, and display operation, with a robust command/response interface for flexible and efficient vision system management.

  • Supported Boards
  • T120F324 Development Kit
  • T120F576 Development Kit
  • Ti60F225 Development Kit
  • Ti180J484 Development Kit

devkit

Edge Vision SoC Framework Block Diagram

Edge Vision SoC Block Diagram

Use Cases

Efinix offers Edge Vision SoC framework that support a wide variety of applications.

Smart Cities

Smart Cities

Traffic Monitoring

Smart Surveillance

Smart Surveillance

People Counting

Robotics

Robotics

Pick-and-Place

ADAS

Automotive (ADAS)

Obstacles Detection


Industrial Automation

Industrial Automation

Defect Inspection

Retail

Retail

Self-checkout Kiosk

Healthcare

Healthcare

Diagnostic Imaging

Algriculture

Agriculture

Crop Monitoring

Advantages of Efinix Edge Vision SoC Framework

Faster Time-to-market

Faster Time-to-Market

With the supported Efinix development kits and ready-to-use software, you can move from setup to a functional prototype in days instead of months. By reducing bring-up complexity and development time, the EVSoC framework enables faster testing and validation, accelerating time to market.

Modular and Extensible

Standard AXI4/APB3 connectivity enables seamless integration of custom compute, control, and AI engines; allowing edge vision systems to be easily customized, expanded, and optimized for evolving application requirements.

Modular and Extensible
Complete Vision Solution

Complete Vision Solution

The EVSoC framework delivers a fully validated end-to-end vision solution—from sensor capture and ISP processing to memory management and display output—simplifying system integration, reducing development effort, and accelerating time to market.

AI Ready Platform

Efinix offers a TinyML platform, accelerate AI inference with the custom TinyML Accelerator on Sapphire SoC suite of RISC-V processors. Built on the EVSoC framework, the platform provides a seamless path to hardware-accelerated inference, enabling developers to add TinyML capabilities without the complexity of a complete system redesign.

AI Ready
Comprehensive Support

Comprehensive Ecosystem Support

Efinix provides function-specific design examples for video processing, AI-powered object detection, machine learning, and multi-camera fusion. With readily available GitHub design examples, comprehensive documentation, and a broad portfolio of Efinix IPs, developers can accelerate innovation, simplify system integration, and bring vision applications to market faster.

AI Ready Platform

The Edge Vision SoC Framework serves as the foundational architecture powering the TinyML platform.

With Efinix TinyML platform, EVSoC evolves into a unified platform for real-time vision and AI processing, enabling:

  • Lightweight object detection
  • Image classification
  • Gesture and activity recognition
  • Sensor fusion and AI pipelines
  • Hardware-accelerated AI inference
  • Real-time vision and AI processing on a single FPGA

From basic ISP processing to advanced AI, the same architecture scales seamlessly to meet evolving application requirements.


A Full Embedded System - Integrates Networking, Storage and Custom RTL

Vision is just the beginning. Many developers need more than vision alone, combining vision processing with Ethernet connectivity, SD card storage, or custom pre-processing IP to build more capable and flexible systems.

Built on the EVSoC framework, the Embedded System Solution extends beyond vision processing with seamless integration of Ethernet streaming with the Triple-Speed Ethernet MAC IP, image and video storage through the SD Host Controller, custom RTL accelerators, and any AXI4/APB3-compatible IP—giving you the flexibility to build application-specific embedded systems.


Resources

Documentation

Edge Vision SoC User Guide

Sapphire RV32 SoC User Guide

Sapphire RV32 SoC Data Sheet

Support Center

Edge Vision SoC (Efinix Edition)

Download from GitHub

Edge Vision SoC (Community Edition)

Open-Source Code

Download open-source source code, example designs, and supporting materials for the Edge Vision SoC framework from GitHub.