Disclaimer: The following recommendations do not prevent thorough testing before deployment of our solution, as any variables can impact VidiReports performances.



Common Minimum System Specs

 

  • Hard drive: 8GB of free space for the software and database

  • GPU: VidiReports does not need a graphic card

  • CPU features: We recommend AVX2 for Core, Footfall and Vehicles. SSE4.2 Is the minimum instruction set we support. 

  • OS:

    • Windows® 7 / Windows® 8 / Windows® 10 / Windows® 11 64-bit.
      Avoid the IOT versions of Windows which may be missing some drivers.

    • Linux Ubuntu 14 or later


Note: Quividi does NOT currently support Android, ChromeOS, Tizen (Samsung) nor WebOS (LG Electronics)


Decentralized deployments - CPU Requirements 


In a "decentralized deployment" mode, VidiReports is running on a local machine that is located next to a screen. This is typically the case when VidiReports runs on the media players that also host the CMS player software.


For new deployments 


  • Indoor deployments
    • Recommended CPU: 
      • Intel I3, I5 or I7 CPU from the 8th generation (CPU model number should be from 8 to 15, e.g. i7-8xxx, i5-9xxx, i311xxx). With such machines, you will be able to run the most recent editions of VidiReports with high camera resolution. 
      • Most ADM processors are also compatible
      • You can also consider most BrightSign players - more info here.
      • Other machines based on Arm Processors are also OK, as long as their processor supports at least the A7 architecture and the NEON instruction set. 
        Note that on those Arm processor machines, VidiReports is only available in the Pro version for Linux (any version) 
    • RAM: 4Gb
    • Camera: any USB webcam supporting the UVC norm (i.e. most mid and high range ones), as well as most IP cameras (in MJPEG or H264 mode). More info here.


  • Outdoor Deployments:
    • Recommended CPU:  If you plan on using vehicles detection, we recommend an Intel I5 or I7 CPU from 8th generation onward.  BrightSign players wouldn't be recommended here.
    • RAM: 8Gb
    • Camera: Outdoor deployment usually require IP cameras (in MJPEG or H264 mode) with an IP66 rating for weather protection. You can also use any USB webcam supporting the UVC norm (i.e. most mid and high range ones).


  • Limited Budget: 
    • Recommended CPU: You can consider machines embedding an entry-level Intel processors from the Intel Celeron range, provided that these Celeron processors come with at least 4 cores, have a and have been released by Intel since 2019 (see a table here). Note that, in that case, you might not be able to detect faces as far as with the i3/i5/i7 machines recommended above. The Celeron N95 or N97 are the good choices for this range (not the N100).
    • RAM: 4Gb
    • Camera: any USB webcam supporting the UVC norm (i.e. most mid and high range ones), as well as most IP cameras (in MJPEG or H264 mode).  More info here.


Note: 

- Windows IoT is not supported out of the box - It's the custom made OS by design, and we cannot test every variation.

- We don't support 32 bits on Intel and ARM processors



For existing deployments


If you already have players in use, please run our benchmark tool and send the results at help@quividi.com. You can find detailed steps in this article.



Centralized deployments - CPU Requirements 


In a "centralized deployment" mode, VidiReports is running on an edge server located in a back-office room and has up to 16 instances simultaneously analyzing video streams provided by IP cameras set on top of points of interest. 


  • Recommenced CPU: We recommend a recent Intel Xeon or AMD Ryzen processor with a large number of cores. As a rule of thumb, you will need 2 Cores per Full HD video stream processed on this server. 
  • RAM: 5Gb of RAM per Full HD video stream processed on this server. 
  • Hard drive: 8GB of free space for the software and database


Additional Ressources:


  • More camera recommendations: Link
  • Best practices for a successful deployment: Link
  • Calculating the maximum detection distance:  Link
  • Balancing Camera resolution and FPS: Link