Image Scanning Help

[Wargot360]

Just a general question to anyone out there that is using the Image Scanning Technology:

What Verdict Ranges and Analysis Sensitivity are people using?

We have set our to the following:

Sensitivity     =     65

Clean            =     0 - 69

Suspect        =     70 - 79

Inappropriate  =     80 - 100

We are finding that we are stopping things that the technology is designed to stop, but we are also stopping alot of business realted images (images inside of presentations) and also alot of things that fall into the personal catergory (baby photos, property listing etc).

I was wondering if changing the Sensitivity would help reduce what the business calls False Positives (both the business related and the personal images), but I am not sure which way I need to go in order to make a difference, and then what changes to the Verdict Ranges I might need to make.

I was just after anyone elses experiance before I started to do some more testing.

[Christopher Smith]

Greetings Ben,

While you can adjust the settings it is typically not recommended to divert from the default settings.

IronPort Image Analysis is intended as an image management  technology.  No real-time image analysis system available today is 100%  accurate and there will always be a level of false positives and  negatives.  

Typical reasons for a false trigger are:

• Small or low resolution images
• Dark or shady lighting
• Distorted colors and surfaces
• Extreme close ups
• IronPort Image Analysis is not current (updated)

If  you have IronPort Image Analysis configured to scan images and are  noticing False Positives or False Negatives, please upgrade to the  latest AsyncOS version for Email to stay updated with the engine  release.

You can submit images which produce false results to the following addresses:

For False Positives: not-inappropriate@imageanalysis.ironport.com
For False Negatives: missed-inappropriate@imageanalysis.ironport.com

IronPort  is committed to further developing its core technology and is always  interested in collating specific images which have produced a false  result. 

Christopher C Smith

CSE

Cisco IronPort Customer Support