add images to database automatically as they are viewed on a thumbnail browser. Since at this time, a cached thumbnail for it already exists.
poll users (on the homepage ? on the ML ?) for the reasons why they use or why they stopped using imgSeek. Also survey future features, where development should head, etc.
a way to search for text on all metadata fields, and not just specific ones
rating system for images, similar to whats done for songs on cymbaline and imms (xmms plugin). Integrated to slideshow viewer and preview pane. Shortcuts for changing image score, and automatic increase on each view. Possibly increasing a value proportional to the time spent viewing it. (Make that an option).
Slideshow would then favor high scored images when picking the next random one. (also an option)
new option on the HTML album wizard: "Preserve image filename", which when enabled would not generate random unique filenames. (To avoid name collisions, i might append random words on the filename) Not sure how useful this can be. Someone please enlighten me.
unify all error msgs output. Right now every error will use "print". Redirect this to a wrapper, which could later be fed to a log or something; In other words, use python's "logging" module. It's available on python2.3. Not sure how much requiring python2.3 or a standalone install of "logging" would piss people off.
write a plugin for other iptc/exif apps. See http://johnst.org/sw/exiftags/
add iptc/exif writeback support
remove distributed distutils module as more people start to support python > 2.2
allow big (256x256), normal (the currently used: proportionally constrained to 128x128), small (64x64) thumbnail sizes. -- shouldn't be too difficult. I could/should still only keep cache for 128x128, as it is already generated anyway when added to content-based db.
make thumbnail cache an option
extend distutils with support for uninstalling. (unlikely. Many have already tried, none succeeded afaik. A distutils rewrite is probably already under work)
forward and backward navigation buttons on html generated pages
Clean callback system and implement event notification better. -- sigh
Visit link:http://www.tasi.ac.uk/advice/delivering/metadata.html#mt9[this page] for more metadata features
Write documentation for XML tags used on the exported files -- does anyone out there even use it ???
code speed up on imgdb.cpp and haar.cpp -- the best way to speed it up right now is a better integration to the underlying image opener api (aka QT or ImageMagick). ImageMagick has some progressive-aware jpeg opener that could be useful when retrieving a thumbnailed low resolution version of the image. Also, imgdb.cpp should work fine with low resolution thumbnails.
transform image before querying (rotate, etc). (a popup somewhere on the import image tab)
more transforms (watermarking, frames, etc) on the Transform dialog (see Transforms class on imgSeekLib/TransformWnd.py) -- bloat. People actually use it ?
ignore images that have too few coefficients. That will avoid web banners, table and html backgrounds, etc.
better debugging, assertion and error msgs on C++ modules
customizable slideshow shortcuts and commands
move files when browsing by System Dirs
Improve performance and usability for huge databases.
Everything else on the "Planned features" section of the site
stuff on http://imgseek.sf.net/todo-code.html
investigate: Good evening, if I start ImgSeek (imgSeek 0.8.2 on KDE 3.2.1 on SuSE 8,2), then the program takes itself more than 90 % of the physical memory, without irgedn something makes. If I implement now actions on the imgSeek data base, then still approx. 70 MT Swap memory is added. With this memory situation naturally no pleasant working more if possible is, can one understand. Which I can do, so that imgSeek becomes faster, except: more memory, whereby I assume that it will not also help, because it seems with imgSeek something to be, but which.