is it possible to download the images of the iconclass dataset with their classfications?
Are there any limitations to this (commercial license and so on)?
Thak you very much
You can download the dataset, which has a large collection of images and their classification codes.
Please let us know how it goes and what kind of interesting things you do with it!
At the moment we would like to understand the possibility to use AI for image classification.
I work fo Cineca (italy). Cineca is one of the most important supercomputing center in Europe.
Good to hear Giorgio. I think it is also very useful for you that Iconclass is also available fully translated into Italian. I do not speak Italian, it would be nice if a native speaker could check some of the searches in Italian - and let us know if it is working OK. Do you mind taking a look?
BTW, I will be attending the VISART VI workshop at ECCV 2022 later this year, where our paper on using some AI techniques for ICONCLASS has been accepted.
It will be very exciting if Cineca is also looking into this.
I went through the Italian translation a bit. I didn’t see anything disturbing for a native Italian speaker.
Perhaps the first letter of the definition (at the first level as in the english version) should be capitalised. In Italian we are used to capitalising only the first letter.
Just an example:
4 · società, civilizzazione, cultura
I would change to
4 · Società, civilizzazione, cultura
I would like to know if, for your paper “The Art Historian’s Bicycle Becomes an E-Bike”, did you a train a model?
In this case is the model freely available?
At the moment we are using the OpenAI model with different text queries.
Thank you very much
We did not train a new model, but used the standard “ViT-B/32” from CLIP.