Hacker News new | past | comments | ask | show | jobs | submit login
Ask HN: Can you recommend any ML papers or blogs for visual art detection?
52 points by rememberlenny on June 16, 2018 | hide | past | favorite | 10 comments



What exactly do you mean by "visual art detection"?

Detecting copies of known images? Detecting styles? Detecting artists? Detecting whether a picture contains art at all?


Check out the work that Local Projects (NYC) did for the Cooper Hewitt Museum. One component of their installation involves museum visitors standing in front of a screen and a computer searching for art that matches their pose.

http://localprojects.com/work/cooper-hewitt-smithsonian-desi...


Look up Plink Art. It was created by a couple of guys at my uni, and eventually acquired by Google a few years ago. They may have a white paper or two kicking around outlining their methodology.


Conceptually, if you’re trying to classify artwork by a category like artist or period, the problem is just ordinary image classification and won’t require anything unique so I’d recommend just reading the current SOA papers for that task.

The foundation for many of the current SOA papers is He, et. al:

https://arxiv.org/abs/1512.03385

(I wrote and maintain Keras-ResNet and I’m always looking for interesting examples. Let me know if you cook up something interesting!)


Lots of people go to art galleries, wondering whether they are really looking at art or not. It seems to be a difficult problem to solve.


You could try something like this: https://www.kaggle.com/c/landmark-retrieval-challenge/discus...

Not art, but a contest-winning approach to a perhaps similar problem of searching a large image database given a photo.


Maybe that google art face detection thing could be a place to start?

https://www.blog.google/topics/arts-culture/exploring-art-th...


Like detecting if there is an art painting in the image, or what era the painting is from?


Would imagine identifying works from stolen art registers via image search would be main use case.


can't think of any off-hand, but a good start would be building a palette histogram for each of the images, I would expect that visual art will have color palettes that are very distinct from photographs.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: