Survey: How many labels you usually assign to the photo?

[Andrey_Ra]

I created this topic to help me clarify some details for my suggestion regarding labels assignment area. You can learn full details of my suggestion on the Mantis: http://mantis.idimager.com/view.php?id=2551.

What I need from you - is the statistics about your personal photo databases. I have prepared special PSU-script which will calculate statistics automatically, you only need to run it through your collections and paste the results here.

Here is the instruction how to run the script:
0. Download script directly from this topic or from the shared dropbox folder (https://www.dropbox.com/sh/7eus2hakoytk ... wl1oa?dl=0).

Label assignments stats_15v2.psc

Main version of the script - all works automatically

(4.5 KiB) Downloaded 696 times

1. Save script in some folder (for example at Desktop).
2. "Tools"->"Scripter".
3. "File"->"Open script (Ctrl+O)"->"Desktop\Label assignments stats_15v2.psc".
4. Run the script.
5. Share the results.
Tip: you can double click on the ?-icon in the Info message to copy the message to clipboard. Then paste it here.

I will collect all submitted results and place them in one graph:

Current Statistics (Average)

The frequency of number of the assigned Labels_Average.png (98.02 KiB) Viewed 14960 times

Current Statistics (Total)

The frequency of number of the assigned Labels_Summarized.png (74.19 KiB) Viewed 14960 times

I wish to thank all who will respond to this request!

[Mike Buckley]

It's nice of you to provide the script, compile all of the data and report back to us.

My results are shown below. It's interesting to learn that 42% of my image files have at least 10 catalog labels assigned to them.

I didn't realize that somehow two images were not assigned any catalog labels. I have now fixed that error. I also see that 12 images are assigned only one catalog label when my labeling convention requires that all images are assigned a minimum of two catalog labels. In fact, images with only three or four catalog labels are suspect. Hopefully I can find a way to determine what those are. My point is that your script is helpful beyond the scope of accumulating the group data.

Info
----
Frequency of number of assigned Labels for selected 10963 images:
0 Labels: 2 ( 0%)
1 Labels: 12 ( 0%)
2 Labels: 280 ( 3%)
3 Labels: 31 ( 0%)
4 Labels: 210 ( 2%)
5 Labels: 456 ( 4%)
6 Labels: 878 ( 8%)
7 Labels: 1318 (12%)
8 Labels: 1546 (14%)
9 Labels: 1494 (14%)
10 Labels: 1298 (12%)
11 Labels: 1093 (10%)
12 Labels: 796 ( 7%)
13 Labels: 588 ( 5%)
14 Labels: 382 ( 3%)
15 Labels: 209 ( 2%)
16+ Labels: 370 ( 3%)

[Andrey_Ra]

Thank you! I have updated the graph.

Mike Buckley's Collection

Frequency_M_Buckley.png (56.02 KiB) Viewed 15321 times

[weidmic]

Here are my numbers :
Frequency of number of assigned Labels for selected 68194 images:
0 Labels: 0 ( 0%)
1 Labels: 4823 ( 7%)
2 Labels: 6006 ( 9%)
3 Labels: 5686 ( 8%)
4 Labels: 8259 (12%)
5 Labels: 13576 (20%)
6 Labels: 15358 (23%)
7 Labels: 7902 (12%)
8 Labels: 5119 ( 8%)
9 Labels: 1141 ( 2%)
10 Labels: 227 ( 0%)
11 Labels: 80 ( 0%)
12 Labels: 13 ( 0%)
13 Labels: 1 ( 0%)
14 Labels: 3 ( 0%)
15 Labels: 0 ( 0%)
16+ Labels: 0 ( 0%)

Cheers

[Andrey_Ra]

Done. Thank you!

Weidmic's Collection

Frequency_weidmic.png (52.58 KiB) Viewed 15321 times

[Hert]

Interesting to see how the average line is flattening out...

[lippe]

Info
----
Frequency of number of assigned Labels for selected 33750 images:
0 Labels: 7081 (21%)
1 Labels: 8617 (26%)
2 Labels: 8900 (26%)
3 Labels: 5138 (15%)
4 Labels: 1938 ( 6%)
5 Labels: 937 ( 3%)
6 Labels: 511 ( 2%)
7 Labels: 252 ( 1%)
8 Labels: 115 ( 0%)
9 Labels: 56 ( 0%)
10 Labels: 72 ( 0%)
11 Labels: 74 ( 0%)
12 Labels: 25 ( 0%)
13 Labels: 18 ( 0%)
14 Labels: 8 ( 0%)
15 Labels: 2 ( 0%)
16+ Labels: 6 ( 0%)

[Ja] [Nee]

I think the hight number of zero labels is due the RAW-JPG are not versioned and only the raw versions have labels.
I notice the Persons and Places category are heavy used. Events are sometime used. Style and Objects category are rarely used.
Lippe,

[Hondor]

Frequency of number of assigned Labels for selected 2831 images:
0 Labels: 98 ( 3%)
1 Labels: 319 (11%)
2 Labels: 826 (29%)
3 Labels: 801 (28%)
4 Labels: 387 (14%)
5 Labels: 213 ( 8%)
6 Labels: 91 ( 3%)
7 Labels: 47 ( 2%)
8 Labels: 24 ( 1%)
9 Labels: 17 ( 1%)
10 Labels: 3 ( 0%)
11 Labels: 0 ( 0%)
12 Labels: 1 ( 0%)
13 Labels: 1 ( 0%)
14 Labels: 2 ( 0%)
15 Labels: 1 ( 0%)
16+ Labels: 0 ( 0%)

[Andrey_Ra]

lippe Thank for your explanations! Zero labels are not included in statistics.
Here is your personal graph:

Lippe's Collection

Frequency_lippe.png (48.32 KiB) Viewed 15321 times

Hondor Thank you too!

Hondor's Collection

Frequency_Hondor.png (49.52 KiB) Viewed 15321 times

Hert
At first I thought about the same . But let wait some time and see what happens next. By the way I prepared second graph with the sum of absolute values. I have not decided how useful this type of statistics can be, but it might be a source of some theories.

[Mike Buckley]

Andrey,

I'm not understanding the absolute numbers. The average on the consolidated graph is much higher than any of the individual numbers. Is that perhaps the total number of images rather than the average number? The absolute numbers on the consolidated graph also aren't consistent with the absolute numbers on each individual's graph, which at least in my case isn't consistent with my data.

Notice that the color representing Lippe's data is not the same in the two graphs, making it difficult to compare his data on the two graphs.

[Mike Buckley]

I think the hight number of zero labels is due the RAW-JPG are not versioned and only the raw versions have labels.

For the data to be meaningful in that situation, my thinking is that you should exclude the JPG files. In other words, I believe the data should be image-specific rather than file-specific.

[Andrey_Ra]

I'm not understanding the absolute numbers. The average on the consolidated graph is much higher than any of the individual numbers. Is that perhaps the total number of images rather than the average number? The absolute numbers on the consolidated graph also aren't consistent with the absolute numbers on each individual's graph, which at least in my case isn't consistent with my data.

Second diagram is represents not averaged values! The yellow line represents summarized data. I did not found any discrepancies - your personal diagram has 1600 on the peak and I see the about the same at the total diagram.

Notice that the color representing Lippe's data is not the same in the two graphs, making it difficult to compare his data on the two graphs.

I have corrected this, thanks!

PS. English is not my native language, so I open to your suggestions of how to name these diagrams to be more understandable for everyone.

[Andrey_Ra]

I think the hight number of zero labels is due the RAW-JPG are not versioned and only the raw versions have labels.

For the data to be meaningful in that situation, my thinking is that you should exclude the JPG files. In other words, I believe the data should be image-specific rather than file-specific.

Oops, at the moment I have placed the new version of the script, that gets all images from the Catalog automatically. Maybe it makes sense to place both scripts - "v2" for simple users, and the "v1" - for advanced?

[Hert]

Looking at the code of your v2 script (I didn't run it yet), I would say that that script gives a more reliable result than the first one. This because when people select versioned image with main images together (for the v1 script) then those labels will be counted double...main and every version. This v2 script on,y looks at main versions which is better.

Ps. I think I see a bug in the v2 script when the progress is cancelled. Either just break and not free there or free and exit instead of break.

[Mike Buckley]

Your English is terrific, Andrey. I'm so envious that you are at least bilingual if not multilingual. I'm an American, so I'm still struggling with English.

By the way, thanks to your script, I identified about ten images that did not have the proper number of catalog labels and corrected the mistakes. Thank you!

Second diagram is represents not averaged values! The yellow line represents summarized data.

The key indicates that the yellow line represents the "average." If you change the key, all will be well. Rather than use the term, "summarized," perhaps use the term, "total." However, I don't know anything about statistics so "summarized" technically might be more accurate.

I did not found any discrepancies - your personal diagram has 1600 on the peak and I see the about the same at the total diagram.

My mistake. The color of the line representing my data is so similar to the color representing Hert's data that I got the two lines confused. I then went brain-dead (in other words, stupid).