Tell me more ×
Cross Validated is a question and answer site for statisticians, data analysts, data miners and data visualization experts. It's 100% free, no registration required.
up vote 2 down vote favorite

I am traditionally a computer vision guy, and have started doing some self-learning in data mining. I'm having a hard time finding places where these 2 fields converge, as I'd like to get a self-study project that relates the two fields.

where are these areas two used together?

share|improve this question
2  
If by data mining you mean something closer to machine learning, then you might look at Sebastian Thrun (used to be at CMU; now at Stanford and Google, I think) and collaborators designing self-driving cars (I think they're now at least partly funded by Google, but it started with some DARPA challenges) and Andrew Ng doing other robotics tasks incorporating computer vision. He also has a nice example of inferring depth (i.e., third dimensional) from 2D still images. There is also a tremendous amount of research by many people in scene recognition from still images. – cardinal Jan 23 '12 at 17:48
I do know of lots of machine learning projects and research going on, but I am thinking moreso towards data mining, and data driven approaches – socks Jan 23 '12 at 17:50
There is also some interesting work of Geoffrey Hinton on automated "natural" image generation via things like Restricted Boltzmann Machines. – cardinal Jan 23 '12 at 17:51
@cardinal: I'd make this an answer. Machine learning is the common thread. socks: do you do computer vision in the abstract, or as a part of robotics? Kalman/particle filters -> time series. – Wayne Mar 25 '12 at 13:30

2 Answers

up vote 1 down vote

Well, there are various places. For example, indexing image data is a natural source of convergence between data mining and computer vision. In order to cluster images, you will probably want to have a tight interplay of computer vision and data mining, too.

Or for example Robust Segmentation of Relevant Regions in Low Depth of Field Images, an image segmentation method that is in fact based on DBSCAN clustering. I bet you can come up with many such examples when you look around. Say: aligning a stack of panorama images requires actually some kind of data mining, too. At least when you have a few thousand SIFT keypoints to work with.

share|improve this answer
up vote 0 down vote

I have seen some work in which researchers are able to harness the availability of large amounts of image data from the web to perform complex tasks (e.g. similarity search).

Two excellent examples I can think of are this paper and the work by James Hays.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.