Hi from Paris,
I’m the archivist for a political and cultural association that’s about to turn a century old.
I’m reconstructing its archives in digital form, classifying photos, videos and documents in directories by date and theme.
Each item is in a single directory.
In total, this collection will contain over 50,000 items in more than 500 directories, for a total size of almost 100 GB.
For the moment, this collection is stored in G-drive.
I’ve just discovered Omeka, which seems to be a very good solution for putting these archives online, and I congratulate its developers.
But before I even start evaluating the Omeka S sanbox, I’d like to ask you the following two questions:
Given its size for a single site, is it possible to keep the collection stored in G-Drive and only put it online through a site built with Omeka?
Is there a plugin that allows automatic facial recognition of 300 or 400 people in photos?
- Or a plugin that lets users propose this facial recognition on each photo by means of a small selection frame and assign a person’s identity to this frame? The proposal would then be validated by a moderator.
Or is this “manual” recognition by selection frame only available to archivists?
Or is the only possibility for archivists to create a tag per person (i.e. several hundred in the long run) and tag each photo?
Thank you for your help.
Best regards
Bernard
I don’t have answers to all of your questions, but I wanted to share some thoughts. You don’t have to upload you files to Omeka to add the metadata and organize them into collections for easy searching.
There is actually a plugin for Omeka Classic that lets you link to or show previews of Google Drive content Omeka Classic - GDriveLinks Plugin . You could create a similar plugin for Omeka S for your site.
I’m not personally aware of a facial recognition module for Omeka. Even if there was, I feel like it wouldn’t work on your images if you don’t upload them to Omeka.
With that said, it does seem like there could be some automation around facial recognition for large image sets, but there could be different levels. For example, there are libraries that will detect facial and create a frame, which that can then be used to manual identify individuals. In order to automatically identify individuals you would need to use a trained data set that knows who you are trying to identify.
Lastly, I work for a company that provides Omeka hosting and consulting. You can add 100 GB+ of files to Omeka. We have several default plans, but if you require additional storage, we can provide that as well.
Hi John,
Thank you for your quick answer.
I understand that I have to combine 1) GDrive for the large storage, 2) Tropy for the management of a large archive of pictures with manual facial recognition and union of pictures from the same item, 3) Omeka S for creation and online access of our website.
That’s seem possible as far as you know ?
Sincerely yours
Bernard
@Bernard , 1 and 3 look good, but I’m not sure I quite understand how you would go about 2 with Tropy and connecting face bounding boxes with people names and then displaying that in Omeka S. If you have a way to to do that, id love to see it.
I’m not as familiar with Tropy so maybe this is possible. I know there is an existing feature request for automatic face detection. If you’re just trying to tag images with people names, and not trying to connect the names with a bounding box, I do think that is possible in Tropy.