incubator-photark-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Subash Chaturanga <>
Subject Re: Documentation of using PhotArk Face Apps
Date Thu, 18 Aug 2011 20:52:26 GMT
Hi all,

The documentation of using Photark face apps has not still changed. So you
can follow the explained procedure as before and train your indexes. So
friend finding can be done either in Facebook or from your private name

- There is a drop down to select from which domain you want to search your
- If you choose Facebook, no need to provide your friends user names,
because the app finds all your Facebook friends and recognizes them.
- If you choose, "Private Name Space", then you have to specify a user name
of an already trained one of your friends (i.e ) NOTE: All
user names must be ended with

Then hit the Recognize button to view your friends in gallery images.

Please try these two apps in your free times, and give your valuable
feedbacks.. :-)

On Tue, Jul 12, 2011 at 1:21 AM, Subash Chaturanga <>wrote:

> *Hi devs,*
> *
> *
> *This is a sample documentation which describes how to use Face
> Recognition functionality.*
> *
> *
> ---------------------------------------------------------------------------------------------------------------------------------------
> Photark Gallery provides a link to visit Face Applications.
> There, you are provided two main apps as options.
>  - Facebook Friend Finder
>  - Generic Face Recognition
> Both these apps are trying to recognize your friends in your PhotArk
> albums.  Facebook Friend Finder finds your friends while  Generic Face
> Recognition app finds friends who are trained in your private name space
> (which is )
> The recognition facility is provided during the album navigation in the
> gallery.
> So before trying to recognize any of your friends from either your private
> namespace domain or from Facebook, you need to tell these apps how your
> friends looks like.
> For that simply, you can train the system by the images of your friends and
> their corresponding names(user names). Lets see how to train PhotArk Face
> App Indexes.
> *- Generic Face Recognition*
> Here, this is related to the friends in your private name space. This can
> be done this by providing an image of your friend as a URL or a local image
> path with a suitable username and a label for the ease of identification.
> The more you train your friend,  the accuracy will be increased.
> i.e Fill the following fields as
> Copy Image File Path -  /home/subash/pictures/adam.jpeg
> Train Username -
> User Label(optional) - Adam Gilly
> And hit the train button. This will update the photark face index with your
> friend. You can train Adam with couple of more of his pictures under his
> username "".
> * - Facebook Friend Finder*
> In this case you don't need to provide images of your friends like in the
> previous app. Because in Facebook, your friends are already tagged in your
> albums. So this app goes through all your friends tagged images and train
> them with their Facebook usernames. You only need to give the userId of your
> friend and system will train your friend. This app doesn't train all your
> friends at startup due to the startup overhead and also you might not want
> to train all your Facebook friends for photark.
> - First click Authenticate Facebook button to provide access of your
> Facebook profile to this app.
> - Then provide the user ids of your facebook friends and train them.
> i.e fill the field
> Train user name - ( or any legal format of your
> friends username)
> And hit the Train button to train. Simple as that.
> -----------------------------------------------------------------------------------------------------------------------------------------
>  --
> Subash Chaturanga
> Department of Computer Science & Engineering
> University of Moratuwa
> Sri Lanka
> Blog -
> Twitter -

Subash Chaturanga
Department of Computer Science & Engineering
University of Moratuwa
Sri Lanka

Blog -
Twitter -

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message