|
|
694aef038b
|
if we use run_ai_on_path, then we remove all existing matches on the path, then go again -- allows us to change threshold and reset
|
2022-02-04 22:03:04 +11:00 |
|
|
|
e1a6348967
|
face_size_limit added so we ignore faces that are too small
|
2022-02-02 21:12:14 +11:00 |
|
|
|
0751cc6010
|
clarified bug-82, fixed issue with AI scan optimising still scanning when it should not, finalised back button work for view/viewlist
|
2022-01-27 21:44:29 +11:00 |
|
|
|
08ca9b4e74
|
partial implementation of first_eid, last_eid -- I think the vals work -- they do for searches anyway, but not stored in pa_user_state yet
|
2022-01-25 00:48:14 +11:00 |
|
|
|
65ebfe2d31
|
DONE - todo to have force rescan delete face from DB
|
2022-01-23 14:54:49 +11:00 |
|
|
|
f8cff5f73e
|
make optimisation from scan do run_ai_on_path IF there are newer reference images than the last scan even if we dont find new files in the scan itself
|
2022-01-19 15:37:27 +11:00 |
|
|
|
2b923f0b60
|
fix BUG that made job mgr break in prod (with DEBUG not even defined, its now False in Prod)
|
2022-01-19 01:20:40 +11:00 |
|
|
|
2af93312e1
|
fix minor syntax, and make sure checkdups waits for getfiledetails not run_ai_on, to avoid the ai job failing due to no ref images, and then withdrawing the checkdups and the ai job cannot create a duplicate
|
2022-01-18 22:11:39 +11:00 |
|
|
|
b327ddff7b
|
optimise to not go past scan job if it does not find any new files, also now print out if we find new files into scan log, fix up first attempt at debug being off/on with "production" or not
|
2022-01-18 22:01:39 +11:00 |
|
|
|
4a1fbf3f44
|
debug off for prod, on for rest by default
|
2022-01-18 21:01:47 +11:00 |
|
|
|
de81db9412
|
unmatched faces now sorts size of face desc, and is slightly prettier -- still slow and only takes you to the file anyway, might optimise that later. still no code to auto deal with unmatched face, but will add some capabilities next. Also, remembered last dir when file_ip/sp/bin. Also throw error if try to find unknown person - happened since I allowed the back/forward.
|
2022-01-18 20:59:39 +11:00 |
|
|
|
096d4f79cf
|
scans now put AI back on by default, and optimise to only call path-level scans once per path, rather than once per dir in path
|
2022-01-16 12:51:25 +11:00 |
|
|
|
a3c75c699d
|
fixed BUGs with deleting old files out of the recyle bin, when they were already deleted from the FS and added timeout debugs, still not 100% the scheduled jobs are working
|
2022-01-15 16:37:40 +11:00 |
|
|
|
bd7cae6037
|
fixed BUG-78 (deleted recursively too far - took out parent_dir), also deleting from the viewer page works with Delete key too. Removed debugs in viewer for fullscreen
|
2022-01-14 13:49:39 +11:00 |
|
|
|
10866c3147
|
scheduled jobs implemented, after X days in settings we scan import, storage paths, and check the Bin for old files and actually delete them from the file system
|
2022-01-13 13:27:55 +11:00 |
|
|
|
a67c20d72b
|
new BUG (very minor), reordered TODOs and now have basic stale job handling - they are detected, and can be cancelled or restarted from GUI
|
2022-01-11 13:18:21 +11:00 |
|
|
|
33ce78cb17
|
fixed Bug-75, moving dir between paths now works
|
2022-01-09 18:35:37 +11:00 |
|
|
|
dc21d65dd7
|
Put new functionality in to allow choosing existing folder in move code - it goes back and forwards 7 days from the date of this file and finds matching files and uses those dirs, or just dirname matches for those dates and offers them up. Also improved Move code to reject dodgy paths
|
2022-01-09 12:20:29 +11:00 |
|
|
|
b1917855e2
|
code clean up, created CreateFSLocation() and refactored code to use it
|
2022-01-08 11:24:53 +11:00 |
|
|
|
665f45c03e
|
working condensed / logical version of moving on Files system - works for multiple scenarios
|
2022-01-07 17:32:23 +11:00 |
|
|
|
ff41164eff
|
over-thought version of new move, about to radically simplify, committing so I dont lost it just in case
|
2022-01-07 15:08:44 +11:00 |
|
|
|
2b0217184f
|
optimise a move via GUI and any re-hashing, due to FS move, or Transform, etc. to update the last_hash_dt so we dont keep redoing the md5 hash as ctime > last_hash_dt, but there is no new content so no new last_hash_dt was being modified
|
2021-10-17 12:05:48 +11:00 |
|
|
|
0ac047f7d6
|
comments
|
2021-10-13 16:53:32 +11:00 |
|
|
|
8f4eb870bb
|
use stat_mtime for gen hash (as moving a file does not change content) AND for fall-back date/time of importing a photo -- Fixes BUG-72
|
2021-10-10 21:12:52 +11:00 |
|
|
|
28a1e73c54
|
log cleanup
|
2021-10-10 12:16:06 +11:00 |
|
|
|
9f2165c04d
|
reset sub dir path properly based on parents path. Also fixed bug-69 - FullPathOnFS() was wrong for first dir in a path
|
2021-10-02 14:48:50 +10:00 |
|
|
|
64466abde2
|
real fix for BUG-64, can now move a directory between import/storage too... Still have move button disabled on selecting a folder in GUI - to be fixed, then we can move 111 Working dir back to the import dir
|
2021-10-02 12:58:45 +10:00 |
|
|
|
3dafae334b
|
remove last remnants of rotate instead of transform, updated rot* svgs to have arrow point in direction of actual rotation, AND, fixed bug where when we rotate an image that has already had an exif_transform on it, then we have to do that and then the requested rotation - fixes why sometimes we rotate say 180 when we intended a 90 rotate. Also, when we transform, recalc md5 as file has changed
|
2021-10-01 22:16:28 +10:00 |
|
|
|
592b283ceb
|
remove debugs
|
2021-09-29 16:56:18 +10:00 |
|
|
|
23c8d16a5b
|
fix BUG-64: can now move files into import or storage path
|
2021-09-26 21:14:08 +10:00 |
|
|
|
92e643f425
|
actually call RemoveDirFromDB when deleting a missing dir, I really screwed this - didnt test properly until now. Should all work finally
|
2021-09-25 13:13:35 +10:00 |
|
|
|
5abd01bb4d
|
fix BUG-66: deleting files out from under and in general CleanUpDirInDB needs to use Entry not Dir
|
2021-09-25 12:11:13 +10:00 |
|
|
|
e9f37aa6a8
|
minor change to optim, if file hash is same but we have no thumb, then try to gen one anyway
|
2021-09-24 20:42:39 +10:00 |
|
|
|
012c2809f1
|
handle case where video is so black we fail to get a thumbnail
|
2021-09-24 20:05:49 +10:00 |
|
|
|
a44810a2f4
|
fix bug with parent_dir being set to an Entry, instead of leaving it being the Dir set earlier in func
|
2021-09-24 17:34:29 +10:00 |
|
|
|
13aab072db
|
use full path for current_file I think its more meaningful, and also extra optimisation - if same md5 hash, dont redo thumbnail
|
2021-09-21 13:46:22 +10:00 |
|
|
|
9f90cdaaac
|
fix up all the dumb bugs I introduced and clearly never tested when I tweaked the remove file/dir after if you move files on the FS out from under PA
|
2021-09-21 00:37:39 +10:00 |
|
|
|
3f21f3d1f5
|
fix not setting in progress when we should on 2 more jobs
|
2021-09-18 17:55:32 +10:00 |
|
|
|
fe7627c89c
|
DEBUG is now a boolean to fix occasionaly debug when I did not want it
|
2021-09-17 18:08:55 +10:00 |
|
|
|
85dd7d7bd1
|
fix bug with commit logs every log, rather than actually every 5 seconds as intentended.
|
2021-09-16 20:22:50 +10:00 |
|
|
|
4d0addb87b
|
removing extra session.commits, and optimsing run_ai_on - dont reprocess faces if no new refimgs since last scan -- so we had set last_ai_scan on FILE as well
|
2021-09-16 19:25:23 +10:00 |
|
|
|
a068133149
|
fixed bug where job is not set to "In Progress" for run_ai_on
|
2021-09-15 18:09:52 +10:00 |
|
|
|
301b3d220b
|
fixed BUG-54/55, viewer now works in PROD, current_file being set in run_ai_on job, added "new" bug-53 re video fname not being set properly when next/prev in viewer
|
2021-09-14 17:56:06 +10:00 |
|
|
|
444b206c64
|
use better face distance sorting/matching algo
|
2021-09-12 18:35:04 +10:00 |
|
|
|
a64b651118
|
use new base_path Setting, but have not tested use of absolute paths instead of relative paths, also need better tooltips for the paths -- AND, still have odd trailing slash due to SymLinkName, etc. being too complex
|
2021-09-05 21:58:54 +10:00 |
|
|
|
eb0145467f
|
fixed bugs 47, 48, 49 and reduced complexity of dup counting as well. Now removes dirs than become empty due to duplicate removals
|
2021-09-04 19:50:49 +10:00 |
|
|
|
6e2d04cd76
|
updated comments
|
2021-08-12 23:22:05 +10:00 |
|
|
|
220200f017
|
face_distance is a float, also pass model through correctly to FaceFileLink
|
2021-07-28 21:41:35 +10:00 |
|
|
|
b7d346c206
|
model_used is now per file, not per face, implemented split of default_model to default_refimg_model and default_scan_model in settings, and default_refimg_model actualy works when creating refimgs in person.py. The model_used in face_file_link is based on default settings model and will scan with cnn if chosen and store that in DB as needed. Need viewer to allow changing per file / not just default for future scans
|
2021-07-27 17:14:03 +10:00 |
|
|
|
cd51ea21bf
|
quick add of default_{model|threshold} to settings, face_refimg_link now stores model_used and face_distance AND working implementation of own face_distance algorithm rather than compareAI(), removed older AI code it would no longer work with DB structures anyway, tweaked viewer to remove coords of unmatched faces for now
|
2021-07-26 18:39:36 +10:00 |
|
