User:The Anome/geolocatable articles that probably already have photographs

This is a small sample of articles which may contain photographic images. Note that the test is currently significantly biased in favour of generating false positives on this list, since they are harmless and will only result in temporarily lost progress which can be made up for in later, more precise, bot scans.

In particular, this shallow analysis will mis-flag many logos, PNG-format locator maps, and coats of arms images as photographs. (For example: at the moment, it looks like identifying File:AlleghanyCountyNC--CranberryTwp.PNG in Cranberry_Township,_Alleghany_County,_North_Carolina as a diagram map may need either a special-case regex or analysis of its image contents.)

I intend to go over this list of articles finding false positives: i.e. articles with no photographs at all that are in this list, and to try in each case to identify why it may have been miscategorized. -- The Anome (talk) 13:21, 22 August 2013 (UTC)

Idea: is it possible to distinguish between PNG photos and PNG diagrams by just measuring file size and area, to give a bits per pixel figure? -- The Anome (talk) 13:42, 22 August 2013 (UTC)

Now added fixes for navboxes, and things like coats of arms and flags included in infoboxes: these are not photographs. -- The Anome (talk) 20:20, 23 August 2013 (UTC)

Now added code which adds better support for checking for re-scaled images, and ignores images inside collapsed boxes, as well as more heuristics to ignore topographic maps etc. -- The Anome (talk) 12:00, 26 August 2013 (UTC)

See also User:The Anome/geolocatable articles that are candidates for photo requests

Reviewed

edit

Not yet reviewed

edit

none