View source for Template:CUDA ISP for NVIDIA Jetson/Main contents

<br>
{| style="border-style: solid; border-width: 1px; margin: 10px;"
|-
| width="100%" valign="center" |
<div style="padding:2rem 1rem; margin:0; background-color:#4d6d9a;color:#ffffff;">  
  <div>
    <p style="font-size: 3em; margin-bottom:0">CUDA ISP for NVIDIA Jetson!</p> 
    <p style="font-size: 1em; margin-top:0; margin-bottom:3rem">CUDA ISP for NVIDIA Jetson Plugin from RidgeRun.</p> 
  </div>
</div>
| valign="center" style="text-align:center;" |
| rowspan="5" valign="top" style="text-align:center;" | {{CUDA ISP for NVIDIA Jetson/TOC}}
<br>
<br>
|- 
| width="100%" valign="top" colspan="2" style="background-color: #4d6d9a; font-weight: bold; text-align: center; color:#ffffff"|
CUDA ISP for NVIDIA Jetson
|-
| width="100%" valign="top" colspan="2"|

{{NVIDIA Preferred Partner logo}}

This wiki is a user guide for our '''CUDA ISP for NVIDIA Jetson''' project. <br>

==What is CUDA ISP for NVIDIA Jetson?==

CUDA ISP is a RidgeRun library developed to provide an out-of-the-box image processing algorithms, focusing on easiness and performance. Since our library's CUDA algorithms run on the GPU, it reduces CPU usage, thus freeing its processing capacity for others applications. This library can also work with computers that do not have a built-in ISP, by adding a discrete GPU to the computer architecture. It is intended for programs that require a large amount of CPU usage.  It can also be used with GStreamer applications.  

The algorithms provided by the CUDA ISP are:

* Demosaic/Debayering
* Auto white balance
* Shifting  
In the image below you can see the software stack of the library. 
|-
| width="100%" valign="top" colspan="3"|
<br>
<br><br>
[[File:SoftwareStackCUDAISP.png|500px|frameless|center|CUDA ISP library ]]
<br>
<br><br>

RidgeRun also makes a binary-only evaluation version available. Please refer to [[CUDA ISP for NVIDIA Jetson/Contact_Us| Contact Us]] to get an evaluation binary.

== Use cases ==

CUDA ISP provides an alternative solution to reduce CPU usage by allowing the GPU handle the image processing. 

* '''Application with a large percentage of CPU usage'''
If you ever found yourself overloading the CPU processing capacity, the GPU can handle it via CUDA ISP.
|-
| width="100%" valign="top" colspan="3"|
<br>
<br><br>
[[File:CPUusageISPv2.png|500px|frameless|center|CUDA ISP library ]]
<br>
<br><br>

* '''Computers with no ISP built-in'''
Computers with x86 architecture do not have a built-in ISP. Adding a discrete GPU to the architecture allows CUDA ISP to be used as an alternative processor.

* '''Artificial intelligence applications'''
Developing artificial intelligence (AI) applications, such as computer vision, can require intensive computations in which the GPU can help with the processing. It is possible to find CUDA ISP working on a multi-GPU environment, where the ISP runs on a GPU and the AI applications on another one.

*'''Streaming applications'''
Streaming applications require lots of CPU processing as well. By letting the GPU handle it with CUDA ISP, the CPU will be freed for other processing tasks.

I am on a Jetson. Why should I use CUDA ISP?

* '''High-CPU usage'''

If you want to free your CPU from the ISP computation, you can use CUDA ISP.

* '''As NVIDIA ISP backup'''

Depending on your processing, the NVIDIA ISP may consume CPU due to additional processing. You can use CUDA ISP as an alternative.

* '''Flexibility'''

CUDA ISP is flexible and extensible. You can add more algorithms to perform ISP tasks. We can team up in your next project to speed up the ISP.

== Tested Platforms ==
Since CUDA ISP is a C++ library, it can be run on different platforms: 
* x86-64 (Linux) with discrete GPU added. 
** Intel-based systems
** AMD-based systems
* ARM 64-bit (Linux)
** NVIDIA Jetson Nano
** NVIDIA Jetson Xavier NX 

== CUDA ISP Purchase ==

{{Shoping Cart - CUDA ISP for NVIDIA Jetson}}

<br>
|-
| width="100%" valign="top" colspan="4" style="background-color: #4d6d9a; font-weight: bold; text-align: center; color:#ffffff"|
RidgeRun Support
|-
| width="100%" valign="top" colspan="4"|
RidgeRun provides support for embedded Linux development for NVIDIA, Xilinx, Freescale/NXP, and Texas Instruments platforms, specializing in multimedia applications. This page contains detailed guides and information on how to get started with the CUDA ISP for NVIDIA Jetson and start using its full capabilities.

To get up-to-speed with your CUDA ISP for NVIDIA Jetson, start by clicking below:
<br>
<br>
[[File:xavier_get_started_here.png|400px|frameless|center|link=CUDA ISP for NVIDIA Jetson/CUDA ISP for NVIDIA Jetson Basics]]
<br>
{{ContactUs}}

<center>
{{Sponsor Button}}
</center>

|}