search

scrubbbbs / cbird

Command-line program for managing a media collection, with focus on Content-Based Image Retrieval (Computer Vision) methods for finding duplicates.
GNU General Public License v2.0
92 stars 5 forks source link

Database changes #21

Open phirestalker opened 2 days ago

phirestalker commented 2 days ago

When i change the default algos on -update for the first time, does it keep those settings for future -update, or do I need to specify each time?

I recently made the mistake of using only dct for some new images I added to the directory. I then tried a new run of update with dct+fdct, but it did nothing. Can you add a way to update the database with new algos?

These are the cases I am thinking and the outcomes I would expect (you may have covered some of these).

  1. I choose only some algos on first index. Subsequent update commands should use the same settings unless otherwise specified.
  2. I specify the algos I want to use It should check the hashes (or whatever) for each file to see if all algos are present. if not, it should add the algos for the files that are missing. This would require checking algos before file existence so that files are not skipped when found on the system (will unfortunately slow down scanning)
  3. I want to remove some algos from the index to prune the size. I would love some options to see what algos are in use for the index currently, and to remove some algos from the index

Also, slightly related. I have some issues with graphical programs started from the command line ICE authority because of Whonix. When it crashes a lot has to be done again. It would be awesome if the matches could be saved to disk while it is processing and able to pick up where it left off. This is especially pertinent for template matching which takes the better part of a day on my lousy system.

I would love to hear your thoughts on this and/or any ideas it might have sparked.

phirestalker commented 2 days ago

I ran -list-index-params on the pictures index and it listed all algos. This means that subsequent calls of update without specifying algos does indeed use them all. As of right now, I guess I will need to start over to get only the algos I want for all images. Once I have a good command with the options I want, I will copy it somewhere to use for every update.

So that gives some more information for the initial question anyway.