Category talk:Computer vision

Suggestions for improvements of the Computer Vision article and category
I would like to start a discussion on how to improve the Computer Vision article and the Computer Vision category.

The following problems need to be taken care of in the article:


 * 1) The current article is a mix between informative text and technical details.  I my view, the presentation is reasonably OK (for now) until and including section 3.  In Section 4, which probably started with good intent as an overview of a typical computer vision system, now seems to be an ever growing list of methods for low-level vision.  This aside, the section appears incomplete, maybe because it is difficult to describe a typical computer vision system after the low-level processing.
 * 2) In section 5, "Related fields", there are only a few sentences about relations to other fields.  This section should either be removed, or I suggest that the last paragraph of section 1 is used as a starting point for a discussion on the relation between Computer Vision, Image Processing, Machine Vision, Pattern Recognition, etc.
 * 3) Section 6, "Applications", appears redundant now that Section 2 provides a rather extensive list of applications.  I suggest to remove it.
 * 4) The "See also" list in Section 7 appears to be a completely random list of articles which have a varying degree of relation to computer vision.  It points both at technology areas such as AI and pattern recognition, to persons with varying degree of relation to the field, to compute vision tasks and to software packages.  The reader sees no apparent structure in this list.
 * 5) The same comment relates more or less to the list of "External links".
 * 6) There are no images whatsoever in this article, which is a bit strange given the topic.

I believe that one problem is that this article appears as a "portal" to the computer vision area, rather than as an informative text about computer vision for someone interested in reading about the area. There is a lack of other places to make a reference to MyFavoriveMethodForSolvingMyFavoriteProblem, and therefore there is a risk that this reference ends up in the Computer Vision article by default. My suggestion is to elaborate on the Computer Vision category page and there define subcategories which can provide more natural places for links to technical matters. For example


 * Category:Computer Vision/Computer Vision Tasks
 * Category:Computer Vision/Computer Vision Software
 * Category:Computer Vision/Computer Vision People

as a start, probably other subcategories may be needed. Some of the more technical stuff which is now in the article could then be moved to the proper category. --KYN 11:34, 5 February 2006 (UTC)

Ok with me
I did the section 3 (structured overview of computer vision tasks). The sub-section about Optical Character Recognition was added by 67.121.68.97, because it doesn't really fit into the existing structure. I just tried to find a classification for typical computer vision problems. My suggestion for a new list of computer vision problems is here:
 * detect presence of known object (for example search engine for images. at least traditional OCR would also fit in here, as it is using bins).
 * determine pose of known object
 * tracking (previous knowledge of pose available)
 * determine ego-position (see here)
 * scene-interpretation (for example mapping by UAVs)

I also did section 4 (structured overview of employed methods). I agree that an exhaustive list of computer vision methods shouldn't be done like this. I suggest that an exhaustive list of computer vision methods should be done by writing articles about them and putting them in an additional category:


 * Category:Computer Vision/Computer Vision Methods

The computer vision article then could retain a section about the typical stages of a computer vision system (image processing, feature extraction, registration).

In my opinion it is very important to provide a concise and highly structured overview of the field of computer vision. However I don't mind in which way this is accomplished. -- Wedesoft 00:34 13 February 2006 (BST)

Rename category to "image analysis"?
I propose to rename this category to "image analysis", instead of "computer vision".

The basis for my proposal is that "computer vision" is a vague term, which gives associations to biology and artificial intelligence. Most computer vision activity is, however, not based on biological or A.I. approaches, but instead use statistical and mathematical approaches to automatic analysis and interpretation of images. Personally, I feel that "image analysis" would therefore be a more neutral term, that better reflects the actual content and subject of this category. --Fredrik Orderud (talk) 20:38, 24 August 2008 (UTC)


 * No. I'm sorry to say this, but I really can't believe that this is coming from you (you wrote list of computer vision conferences!!). I recommend a second reading of computer vision. --Jiuguang (talk) 21:31, 24 August 2008 (UTC)