up vote 24 down vote favorite
4

Proposals: Computer Vision, Signal, Image & Video Processing

Shall we merge Computer Vision and Signal, Image & Video Processing? As computer vision is an application of signal processing, it would make perfect sense.

link
I asked a question about pending merges on meta: meta.stackoverflow.com/questions/94578/… – Traroth Jun 13 '11 at 12:46
1  
So, Dignal Processing is now in beta stage, and was not merged with Computer Vision yet. Merge decisions seem to be taken randomly. How disappointing... :-( – Traroth Aug 18 '11 at 14:59
So this merge proposal is going nowhere, and in the meantime, the Signal Processing beta is not doing well, either, while Computer Vision is stuck in commitment phase. So instead of a healthy site, we could end with... nothing at all! Everything's fine! – Traroth Mar 12 at 14:35
feedback

4 Answers

up vote 17 down vote

Considering the current situation (as of June 15, 2011), The Computer Vision proposal maxxed out at about 40 committers. Yes, a lot of vivid discussions and enthusiasm, but just 40 users. The SigProc.SE proposal gets a steady stream of committers and is well poised to a beta.

In other words, the CV proposal don't seem to be viable. If they need a place for Q&A, they'll need to consider using SignalProcessing.SE.

Some statistics, as of June 15, 2011:

  • Among Computer Vision committers (43 total), 27.9% also committed to Signal Processing.
  • Among Signal Processing committers (157 total), 7.6% also committed to Computer Vision.

That is, 12 of us committed to both. Ideally, we should try to get at least half of the 43 CV committers to also commit to Signal Processing, if we want the merge to move forward.

Edited: Adding a bit of historical perspective.

  • During the definition phase (on-topic and off-topic questions), at some point there were about 110 followers.

CV people usually know how to partition their questions into SignalProcessing and Artificial Intelligence (Machine Learning) parts.

In addition, they may ask some Higher-dimensional Signal Processing (dim >= 2) questions which require some theoretical / practical knowledge not found in 1-D signal processing. (For example, 2D/3D geometry.) Also, CV are equally interested in Frequency domain and Spatial domain, as opposed to audio processing which tend to heavily focused in frequency domain. If the 1-D signal processing crowd at SigProc.SE are able to tolerate higher-D questions, then CV people will be comfortable asking questions at SigProc.SE.

A problem of the Computer Vision proposal was that a lot of interested people have long since left the academia (going to work after graduation), and usually they find themselves in a rather lonely situation (the only vision guy in the office, with everyone else working on other types of programming), and they won't be able to recruit a lot of friends and acquaintances to join the CV.SE proposal.

Those still in the academia (college students, graduate students, researchers and professors) have a lot of recruiting power. (One class of Computer Vision students * Class taught once per year * number of colleges offering this class >= much greater than 40) Yet there are some reasons they don't come to the CV.SE proposal in flocks. It may take some community insight to find out why this is the case.

(a lot of assumptions and biases from me - please correct me if i'm wrong.)

link
feedback
up vote 7 down vote

I think they would work rather well together. Also, both really need people to commit, so this might boost the process of recruiting. I'm not sure if the Computer Vision crowd classifies itself as an artificial intelligence community more than an image processing community, but I think there's definitely enough overlap.

link
feedback
up vote 6 down vote

I think merging the Computer Vision and the Signal, Image & Video Processing proposals is a good idea. How do we do it?

link
Vote for this question if you didn't already. – Traroth Aug 18 '11 at 15:01
feedback
up vote 1 down vote

Actually, maybe Computer Vision should be merged with Image & Video Processing and Signal Processing should remain as a stand alone topic.

link
1  
I think the people who started Signal, Image & Video Processing will not be ok with that. Questions about how to process images and videos using DSP should be on-topic. – Traroth Jun 22 '11 at 14:05
@Traroth: as long as the existing users of Signal Processing do not mind the abundance of non-DSP (non-frequency-domain) techniques used solely in image processing (2D/3D). Some part of computer vision/image processing has more to do with Photography, human cognition (machine learning, ad-hoc algorithms inspired by human visual model, or some other hard-coded bizarre ideas), etc. – rwong Jun 24 '11 at 3:28
@rwong: Sure. I don't think the Signal, Image & Video Processing proposal is meant to be only about DSP-related Q&A anyway. If you look at the site description and the on-topic questions, all image processing questions are on-topic, for example. – Traroth Jun 24 '11 at 8:58
1  
@rwong: DSP is not frequency-domain only. I'm not sure where you get that from. All the things you mention in your comment are of interest to the people I see interested in Signal, Image, and Video Processing. – Peter K. Jun 25 '11 at 19:19
@Traroth: Should have responded more directly. – Peter K. Jun 25 '11 at 19:19
feedback

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged