I think we need to break down your question...

It seems that the thematic folders and the file names may be ok 
descriptive tag sources to start with.  Perhaps you could try to 
identify patterns to extract information for tags (i.e., "hall", 
"committee", "holiday", etc.) You could traverse the file system, and 
use the Google Data API for Picasa, to do an initial upload with tags 
generated from those folders and file name patterns.  You will need 
*some* kind of user input to get to more detail.

So, are you asking whether it's possible for software to generate 
metadata for what is pictured in a photograph without user input?

Are you intending for facial recognition software to identify who is in 
a picture or just that there are people in it?

I think you're on the right track by using Picasa.  I think it may be 
easier for people in those departmental units to help tag people, and 
add descriptions through that interface.  Then you can write some kind 
of script that slurps them into your repository.

Are you using the Google Data API for Picasa?

For batching jobs, you might look at GoogleCL:

If you don't want to or are not allowed to upload all your photos to the 
web first, you will probably have to look at Pyfaces 
[] or Open Source Computer Vision 
[], but they smell of 
pain to me.


On 11/20/12 2:54 PM, Kyle Banerjee wrote:
> I am in the process of examining how photo collections maintained by campus
> units can be incorporated into the library's repository. In all cases that
> I've had to deal with so far, they're just using the file system -- i.e.
> traversing folders that arrange images thematically to file names that
> indicate the content.
> Each of these collections contains many thousands of images. This means
> that it's a hassle for them to find images, but also that there's no way
> library staff alone will be able to handle all the metadata creation.
> I'd like to use something slick like picasa to help them out (facial
> recognition is an especially big deal for us). But I'm finding the metadata
> to be both minimalist and clunky to work with so I wanted a reality check
> to see if I'm not doing this the dumb way. Things I've noticed:
>     1. Picasa appears to store info in xmp rather than exif which is great
>     given the limitations of exif. However, I haven't yet found a way to use
>     more than a couple fields. The caption shows up in a description field, and
>     they tag show up in subjects. But aside from that, I'm at a loss of how to
>     populate other DC fields through the interface.
>     2. Facial recognition metadata doesn't show up in xmp at all. However, I
>     can get that by parsing .picasa.ini and contacts.xml (clunky, but doable).
>     I'm kind of tempted to tell people to go into albums and batch tag the
>     people albums since it's going to be fun explaining how to locate these
>     hidden files.
> My real question is whether anyone has come up with a really good way to
> assign metadata to thousands of photos, preferably in batch fashion? Thanks,
> kyle

Shaun D. Ellis
Digital Library Interface Developer
Firestone Library, Princeton University
voice: 609.258.1698 | [log in to unmask]