Getting Started

Introduction

CUVI is an accelerated ‘Computer Vision & Imaging Software Library’ that leverages many-core hardware of GPUs to add acceleration to your existing vision and imaging applications as well as provides essential functionality to build a solution from scratch. CUVI comes in easy to use, off-the-shelf, modular interface that takes minimal time and developing effort to integrate with your existing applications. CUVI performs magnitudes faster than competing solutions and requires no GPU-programming experience on the part of user.

Download

You can download the latest trial version of CUVI from the website. You'll be asked to signup for CUVI updates after which you will receive download link for Windows OS. Free downloads include Windows OS support for only NVIDIA GPUs. The image/video size limit in free version is 1000x1000 pixels. You may contact us from the website's contact form or by sending an email on contact@cuvilib.com to purchase CUVI licenses from professional usage and integration support.

Windows Installation

Requirements

• CUDA capable Graphics Card [see list]
• Latest drivers from NVIDIA [download]
• Microsoft Windows 7 or above.

Install

Extract cuvi package to any location on your machine. The package contains CUVI binaries and all required and supporting libraries. Add the bin folder into System path. If you are using licensed version of CUVI, add the system variable named 'CUVI_PATH' pointing to the location of license.dat file, for example, if the license file is located at D:\CUVI\1.6.5\license.dat then add variable: CUVI_PATH value: D:\CUVI\1.6.5\ (Note the trailing '\') Note: The trial version now doesn't need a license. If everything is correct, you should be able to run example.exe located in test\ folder and see the results right away. For support, you can contact us at contact@cuvilib.com

Visual Studio Integration

Using CUVI in your existing Visual Studio project is as easy as just creating a VS project. CUVI package comes with library file and an include folder that you can add to your existing or new project. To get you started, CUVI package contains several source cpp files to help you get used to the CUVI I/O and framework. For further information on how to use CUVI features, you can visit the following links:
- Image I/O & Framework
- CUVI by Example
- CUVI Features

Linux Installation

Requirements

• CUDA capable Graphics Card [see list]
• CUDA Tooklit version 10.2 [download]

Install

• Inflate CUVI package to any location on your machine.
• Run install.sh or manually copy all the lib and include folders into usr/local/lib and usr/local/include respectively,
• If you are using licensed version of CUVI, put the license.dat file into home/ directory.