Author Topic: Images classification  (Read 6340 times)

rosko

  • Homo erectus manualfocus
  • NG Supporter
  • **
  • Posts: 1317
  • France/Uk
Images classification
« on: December 09, 2016, 14:41:22 »
Images classification.

So far, I use to classify my pictures by year ► general subject ► subfolder ► and file.

For instance,

# 2016 ► NATURE ► FAUNA ► birds
                                              ► insects 
                                              ► other
                             
                               ► FLORA ► plants
                                               ► muchrooms/algaes/lichens

           

             ► LIFE      ► HUMAN ► steet
                                               ► portrait

                              ► TRAVEL► France
                                               ► Abroad


I think that most of us adopt this classification.

However, I am planning to change it next year, classify them by Lens first. Just because I noticed that I often want to find a picture with my 125mm or my 55mm for exemple. and it's not easy to find.

Is anybody here doing this classication by lens already ? And are-you happy with it ?

Thanks to meet this discussion.

Francis.

Francis Devrainne

Bjørn Rørslett

  • Fierce Bear of the North
  • Administrator
  • ***
  • Posts: 8252
  • Oslo, Norway
Re: Images classification
« Reply #1 on: December 09, 2016, 15:00:12 »
Why not build a relational data base instead? Then you can retrieve images on many criteria, also focal lengths if you so prefer. Just include relevant EXIF fields. The data base will just hold pointers to the images themselves, thus you can continue to store images in whatever filing system you find useful.

If you have GPS coordinates to the images you can even do reverse map lookups to find images taken in a certain area. That is perhaps the data base feature I use most frequently myself.

MFloyd

  • NG Member
  • *
  • Posts: 1801
  • My quest for the "perfect" speed blur
    • Adobe Portfolio
Re: Images classification
« Reply #2 on: December 09, 2016, 18:57:48 »
Relational database is the way to go. A hierarchical database will very soon show his limits and become a data black hole if by any chance  something lands in the wrong place.
Γνῶθι σεαυτόν

Andrea B.

  • Technical Adviser
  • *
  • Posts: 1671
Re: Images classification
« Reply #3 on: December 09, 2016, 20:21:09 »
Building a databsase is rather a lot of work when we have nice search tools which permit searches of existing EXIF to find all images having certain settings.

Also in apps like Photo Mechanic using the IPTC stationary all images can be assigned fields of identifying data and also keywords during ingest and naming. These fields and keywords can later be searched on by Photo Mechanic or other IPTC tools.

Bjørn Rørslett

  • Fierce Bear of the North
  • Administrator
  • ***
  • Posts: 8252
  • Oslo, Norway
Re: Images classification
« Reply #4 on: December 09, 2016, 20:55:02 »
It's also a question of scalability. Any approach needing to read through all files to retrieve information will sooner or later be cumbersome and too slow, unless indexes are built and used for searching. Thus even when a full-blown relational data base might be deemed overkill, it has its merits and some kind of hybrid solution at least should be considered. I employ both variants.

Andrea B.

  • Technical Adviser
  • *
  • Posts: 1671
Re: Images classification
« Reply #5 on: December 09, 2016, 21:35:24 »
What DB software are you using?

MFloyd

  • NG Member
  • *
  • Posts: 1801
  • My quest for the "perfect" speed blur
    • Adobe Portfolio
Re: Images classification
« Reply #6 on: December 09, 2016, 22:28:33 »
I'm simply using Lightroom (Lr) which LrCat (catalog) seems to be a database build around SQLite. More of this here: http://www.earthboundlight.com/phototips/querying-sqlite-lightroom-database.html

From my experience searching the LrCat within the Lr application is fast and efficient.
Γνῶθι σεαυτόν

Airy

  • NG Supporter
  • **
  • Posts: 2701
    • My pics repository
Re: Images classification
« Reply #7 on: December 09, 2016, 22:38:52 »
Multiple tagging (in LR) does the job for me. I created several tag trees (optics; lenses; subjects; places; people; etc.). The only trouble is, it ties me to LR.
Airy Magnien

HCS

  • NG Supporter
  • **
  • Posts: 1533
  • The Netherlands
Re: Images classification
« Reply #8 on: December 10, 2016, 10:40:39 »
Building a databsase is rather a lot of work  ...

While that is obviously true, if you don't do it, the files will get lost. It's not a matter of if, but when.

Many a photo database program can harvest the technical data so you can use that already, while it lets the files be where they are. In this way, the actual tagging or keywording can be done at ones own pace and spread over time for the existing image base, while immediately doing it for new acquisitions.

I use Photo Supreme (PostgeSQL version) for this, but there are many like it. When you would try PSU, don't be fooled by it deceptively simple look when first opening it. Most of its features are driven by configuration.
Hans Cremers

Øivind Tøien

  • NG Supporter
  • **
  • Posts: 1890
  • Fairbanks, Alaska
Re: Images classification
« Reply #9 on: December 10, 2016, 12:06:40 »
I have dutifully been adding keywords to my Nef files in ViewNX, only missing the first two years of my DSLR area. I have no category system or any hierarchy, I just add keywords as I see fit. (I really need something simple to keep up). But admittedly I have not got a good way of searching these (!). For a while I experimented with using Windows Explorer that had some capabilities including searching for IPTC keywords in .Jpgs (but not ratings); somehow that seems to have been lost in Win 7 or I have not got it configured right. I strongly prefer that the keyword information etc. reside with the image file, not in a central database. Any program used for searching should of course be able to index this information (as also Explorer did), but the original information should reside with the file. I have my files simply chronologically ordered, as that is the easiest way to remember them for me.

The keyword and other IPTC managment like star ratings has been very easy up to now, with ViewNX including the IPTC in the NEF. I wonder how things would work for me with sidecars. My ViewNXi and CNX-D testing so far indicates that if previously added in the nef the IPTC information will travel to the sidecar first time it is edited, however after that any edits done to keywords or ratings will not be written back to the NEF. During my initial testing of CNX-D and a number of other programs I did not get the impression that these Xmp files were very compatible between different applications. These issues are particularly critical with ratings, as a lot of work goes into these, and ratings often need to be changed after an image has been edited (and sometimes I also forget to add keyword/copyrights info and have to do it post edit).

I like to hear more about applications used for image search, although the original thread really was about how it is organized.
Øivind Tøien

simato73

  • NG Supporter
  • **
  • Posts: 1128
  • You ARE NikonGear
Re: Images classification
« Reply #10 on: December 10, 2016, 12:28:21 »
While that is obviously true, if you don't do it, the files will get lost. It's not a matter of if, but when.

Many a photo database program can harvest the technical data so you can use that already, while it lets the files be where they are. In this way, the actual tagging or keywording can be done at ones own pace and spread over time for the existing image base, while immediately doing it for new acquisitions.

I use Photo Supreme (PostgeSQL version) for this, but there are many like it. When you would try PSU, don't be fooled by it deceptively simple look when first opening it. Most of its features are driven by configuration.

Photo Supreme is one of few single user friendly full fledged digital asset management options.
I am in the process of switching to proper digital asset management from Aperture (which in the last few years I was using only for DAM, not for editing) and PSu is one that I am trialling, currently in my second week of the trial.
I am also trialing (from yesterday) another software from Norway called Fotostation.

I have done a bit of research on this theme recently and have found some interesting blog posts that others might find interesting:

http://www.bkwinephotography.com/technology/good-dam-digital-asset-management-software/

http://www.bkwinephotography.com/technology/features-functions-digital-asset-management-system-dam/

http://www.bkwinephotography.com/technology/found-best-digital-asset-management-dam-system/

Some of the options are not available for me because I am using a Mac.
For example the Windows crowd could try Daminion and for the Linux crowd there is digiKam (the latter technically has also an OS X version but it crashes all the time; pity because it is free and open source with a vibrant community of developers)
Simone Tomasi

rosko

  • Homo erectus manualfocus
  • NG Supporter
  • **
  • Posts: 1317
  • France/Uk
Re: Images classification
« Reply #11 on: December 10, 2016, 13:11:25 »
Why not build a relational data base instead?

Thanks everybody for all suggestions.

A relational database is too complicated for me. I am not expert enough to use these high level systems. I will find out among software as suggested here or make an additional classification like LENS ►subject►file.

If I look at  500mm to ''nature'', that will be obviously mostly birds.

If I look at 55mm macro, that will be mostly insects or flowers...

In addition, I can save them on separate CDs which is one more chance to keep my pictures safe.
Francis Devrainne

Bjørn Rørslett

  • Fierce Bear of the North
  • Administrator
  • ***
  • Posts: 8252
  • Oslo, Norway
Re: Images classification
« Reply #12 on: December 10, 2016, 13:28:57 »
I think what Andrea had in mind is the building of the support structure for a given data base. The DB itself can obviously be of the 'canned' variety such as SQLite, mySQL, etc. so no need to start from scratch in that regard.

Anthony

  • NG Supporter
  • **
  • Posts: 1619
Re: Images classification
« Reply #13 on: December 11, 2016, 13:28:39 »
I use Neofinder. http://www.cdfinder.de/en/en/photos.html

It is very fast, simple and not expensive.
Anthony Macaulay

Andrea B.

  • Technical Adviser
  • *
  • Posts: 1671
Re: Images classification
« Reply #14 on: December 11, 2016, 18:19:45 »
The NeoFinder tool looks quite nice, too.
I want to look into the various other referred tools.




Two things to deal with: 
  • adding IPTC metadata to the EXIF on a per-file basis
  • storage and search
Adding IPTC metadata to the existing EXIF in each image file is important so that files may be sent for publication or other usage with intrinsic copyright, location and other relevant identification attached. And also important in the case where an image file becomes "separated" from its storage. (I can't tell you how many times I've found "stray" files which have been misdirected during a Save As.)
If the IPTC metadata is in place, then it becomes easy to organize or reorganize your image files. And during ingest with the right app or tool, adding initial IPTC metadata is easy. It can always be refined or added to later.

Of course - being the lazy person that I am - rest assured that I have hundreds of image files which have no IPTC data yet added.  ;D This in spite of having worked up and stored good IPTC templates in Photo Ninja.

I need a photo management assistant.  8)