Software recommendation for image tagging?
February 3, 2018 5:19 PM Subscribe
I'm looking for a way to speed up what is currently a tedious data entry project. I have a set of approximately 700 photos. Each photo can be of up to 8 items (e.g. cat, dog, house, cup...) of a set of 20 total possible items. I need to tag each photo with the list of objects in each photo.
Rather than type the name of each item repeatedly I was wondering if there was some software that would let me see each image, and click "cat" "dog" and "cup" from a list of the 20 possible items. Then I need a .csv/Excel file containing the file name, and the list of items. (e.g. photo_123.jpg, contains: cat,dog,cup) Any ideas would be very welcome. Thanks!
Rather than type the name of each item repeatedly I was wondering if there was some software that would let me see each image, and click "cat" "dog" and "cup" from a list of the 20 possible items. Then I need a .csv/Excel file containing the file name, and the list of items. (e.g. photo_123.jpg, contains: cat,dog,cup) Any ideas would be very welcome. Thanks!
I've been using Crowdflower for tagging images like this. You can upload a csv of the images and the number of categories. You can have other people tag them, or get a link to send to your own 'internal team' (which might just be you).
posted by todolos at 6:24 PM on February 3, 2018
posted by todolos at 6:24 PM on February 3, 2018
I would approach this using a simple one table database.
- From the command line list all the filenames, redirect that list into a file.
- Import the file into a new database table.
- Add columns to the table for each of the items - boolean Yes/No values defaulting to False.
You can then scroll through the list side by side with the photos displayed in finder (or your favourite image viewer) and tick the boxes for each item.
When finished you can export/copy back to Excel.
Many database systems will allow you to import the images right into the database, but I think that's overcomplicating things for this.
posted by Lanark at 1:48 AM on February 4, 2018
- From the command line list all the filenames, redirect that list into a file.
- Import the file into a new database table.
- Add columns to the table for each of the items - boolean Yes/No values defaulting to False.
You can then scroll through the list side by side with the photos displayed in finder (or your favourite image viewer) and tick the boxes for each item.
When finished you can export/copy back to Excel.
Many database systems will allow you to import the images right into the database, but I think that's overcomplicating things for this.
posted by Lanark at 1:48 AM on February 4, 2018
If you find this let me know. :) I once wrote a script that did something similar but it's unlikely I still have it (I'll post if I find it).
Just a couple of notes that may, or may not, help:
* you can put images in excel, though I don't know what the process for adding 700 would be
* Alternatives to Crowdflower would probably include Amazon Mechanical Turk and Fiverr.
* That irfanview article that WCityMike posted looks promising
posted by Awfki at 8:39 AM on February 4, 2018
Just a couple of notes that may, or may not, help:
* you can put images in excel, though I don't know what the process for adding 700 would be
* Alternatives to Crowdflower would probably include Amazon Mechanical Turk and Fiverr.
* That irfanview article that WCityMike posted looks promising
posted by Awfki at 8:39 AM on February 4, 2018
You should be able to do this in Adobe Lightroom. Tags are called keywords. Adding keywords gives you a clickable list, and you can batch keyword photos in library mode (a grid in which you see multiple photos) if desired. I have an older version but I think the new one automatically recognizes objects in images so you might be able to search for "dog", tag those as a batch, and then clean up later. You'll need a plugin to export like this one.
posted by beyond_pink at 10:17 PM on February 4, 2018
posted by beyond_pink at 10:17 PM on February 4, 2018
This thread is closed to new comments.
posted by WCityMike at 6:16 PM on February 3, 2018