SAFR Server requires at least 5 GB of GPU RAM but ideally a GPU wiht 6 GB RAM.  If video feeds are being processed on the server also, you will also need about 700 to 900 GB RAM pers video feed.  


If your system does not have sufficient GPU RAM, you should attempt to upgrade the GPU.  If that is not possible, following may be possible workarounds

  • Install SAFR Server without age, gender and sentiment models
  • If Mask recognition is not needed, also omit mask, mask-recognition and occlusion models
  • If possible, more camera feed processing to separate PC.  


Omitting age, geneder and other models can only be done by installing server via command line - see SAFR_Platform_Command_Line_Install_Options


  1. Download SAFR Platform installer

  2. Run the installer from command line with the flags needed to disable the models. 


Here is the command you would run (the filename needs to be changed to the name of the file you downloaded):

SAFRPlatform_win_1_8_302_08_13_19.exe /Age=NO /Gender=NO /Sentiment=NO


If you feel you do not need mask recognition, you can also run the following instead which will free up even more GPU memory:

SAFRPlatform_win_1_8_302_08_13_19.exe /Age=NO /Gender=NO /Sentiment=NO /Mask=NO /MaskIdentity=NO /Occlusion=NO