[guardian-dev] Super Image Resolution Impact on CameraV

[Nathan of Guardian]

Thanks for the note. Regarding face blurring, I think you are thinking
of ObscuraCam, and not CameraV (which is an encrypted camera for
evidence gathering).

ObscuraCam in fact does offer a full pixel removal "black bar" option.
In addition, we don't blur the photos, but actually pixelate them using
a technique that is resistant to the kind of reconstruction you mention.
 You can find some of the native code we use here:
https://github.com/guardianproject/ObscuraCam/blob/master/app/src/main/jni/redaction.h

On Thu, Sep 1, 2016, at 08:58 PM, bancfc at openmailbox.org wrote:
> I came across something which I think impacts the security assumptions 
> of CameraV.
> 
> Super image resolution is becoming a reality meaning that very blurred 
> face pics can be reconstructed with great accuracy because of machine 
> learning. I think the only way around this is to blot out faces 
> completely with solid colors.
> 
> https://github.com/david-gpu/srez/blob/master/README.md
> 
> 
> 
> _______________________________________________
> List info: https://lists.mayfirst.org/mailman/listinfo/guardian-dev
> To unsubscribe, email:  guardian-dev-unsubscribe at lists.mayfirst.org


-- 
  Nathan of Guardian
  nathan at guardianproject.info