552 Commits

Author SHA1 Message Date
Sefik Ilkin Serengil
5825e2aa88 GhostFace model added 2024-03-16 09:41:08 +00:00
Sefik Ilkin Serengil
33f4d2c314 todo removed 2024-03-16 08:47:15 +00:00
Sefik Ilkin Serengil
20b990ac07 streaming code refactored 2024-03-16 08:39:01 +00:00
Sefik Ilkin Serengil
820ba5946b resolving issue 1098 2024-03-11 18:39:00 +00:00
Sefik Ilkin Serengil
ec4c3cd17c resolving issue 1099 2024-03-11 18:36:30 +00:00
Sefik Ilkin Serengil
9705e6f4ef resolving issue 1097 2024-03-11 18:31:52 +00:00
Sefik Ilkin Serengil
049ab8cdff resolving bug 1096 2024-03-11 18:29:52 +00:00
Sefik Ilkin Serengil
7753476978 layer for external models are no longer being used 2024-03-10 15:49:48 +00:00
Sefik Ilkin Serengil
e13f3cf8ad simlipfy yolo result is valid 2024-03-10 10:50:39 +00:00
Sefik Ilkin Serengil
e67614ff7a exception handling in case of yolo has no keypoints 2024-03-10 09:08:29 +00:00
Sefik Ilkin Serengil
42e911958b tf requirement for deepface package 2024-03-09 22:34:55 +00:00
Sefik Ilkin Serengil
ca4bd256d9 docstring updated
in verify, we mentioned similarity metric but it should be distance metric
2024-03-09 21:06:25 +00:00
Sefik Ilkin Serengil
65c47badb5 LocallyConnected2D not supported after tf 2.12
- this requirements is just for deepface model. so, instead of enforcing lower tf versions
we should check the current tf version in deepface model initializaiton.
2024-03-09 21:05:19 +00:00
Sefik Ilkin Serengil
54d678ca41 get rid of unused import, add a comment 2024-03-08 23:14:17 +00:00
Sefik Ilkin Serengil
392160f6fc hash with properties instead of file itself 2024-03-08 23:09:10 +00:00
Sefik Ilkin Serengil
4b3473e023 docstring for pre-calculated embedding support in verify 2024-03-08 20:22:22 +00:00
Sefik Ilkin Serengil
259add4a13 simplify getting pickled images 2024-03-08 19:52:38 +00:00
Sefik Ilkin Serengil
6eced68e69 support embedding input for verify 2024-03-08 15:22:58 +00:00
Sefik Ilkin Serengil
d7c2998e1b wrapper for find distance added 2024-03-08 13:55:02 +00:00
Sefik Ilkin Serengil
2f9f9761d0 cover uppercase links 2024-03-08 12:01:45 +00:00
Sefik Ilkin Serengil
07a2d5bf62 enhancement for find function
- detect replaced files already in data store
- store column names in the pickle
2024-03-08 12:01:31 +00:00
Sefik Ilkin Serengil
20f916c024 distance as float instead of numpy float 2024-03-02 20:43:47 +00:00
Sefik Ilkin Serengil
29ffce8594 types for eyes 2024-03-02 20:43:16 +00:00
Sefik Ilkin Serengil
e70496ba6f avoid multiple feature extraction for faces in image pairs 2024-03-02 20:20:33 +00:00
Sefik Ilkin Serengil
715d6317d4 image loading steps refactored
- if an image starts with http, then it was trying to download from web
- to make base64 loading compatible with find, now jpg jpeg png images are accepted.
2024-03-02 12:39:07 +00:00
Sefik Ilkin Serengil
50dd5d7aa2 fastmtcnn sometimes fails 2024-03-02 12:26:02 +00:00
Sefik Ilkin Serengil
a9cb7b96e3 throw meaningful error if an img_path cannot be read 2024-03-02 11:57:21 +00:00
Sefik Ilkin Serengil
d3f7cab119 new package version 2024-03-02 11:56:43 +00:00
Sefik Ilkin Serengil
87f4893205
Merge pull request #1047 from AndreaLanfranchi/al20220227-rotate-facial-area
Simplify and strengthen rotate_facial_area
2024-02-27 11:07:52 +00:00
Andrea Lanfranchi
90053295cb
Change expression for direction detection 2024-02-27 12:00:47 +01:00
Andrea Lanfranchi
d0ffc1bfde
Amend quirk behavior of mod operator 2024-02-27 11:31:41 +01:00
Andrea Lanfranchi
923f2cfb4c
Remove unwanted comments 2024-02-27 10:58:39 +01:00
Andrea Lanfranchi
08710029e8
Remove commented out code 2024-02-27 10:52:10 +01:00
Andrea Lanfranchi
b62d6cc9a0
Amend wrong calculation of increased w and h 2024-02-27 10:51:13 +01:00
Andrea Lanfranchi
fe077308f4
Remove commented and unused code 2024-02-27 10:31:24 +01:00
Andrea Lanfranchi
fb6d2f572e
Amend double increase of expand_percentage 2024-02-27 10:29:15 +01:00
Andrea Lanfranchi
53510f0584
Linting 2024-02-27 10:15:50 +01:00
Andrea Lanfranchi
001e609a82
Simplify and strengthen rotate_facial_area 2024-02-27 10:02:38 +01:00
Dries Verachtert
07e695924b Correct info about returned types for the represent method 2024-02-26 16:44:45 +01:00
Sefik Ilkin Serengil
7ee55cb682 common confidence scales 2024-02-25 14:25:17 +00:00
Dries Verachtert
7e14fc875d Correct info about returned types for extract_faces 2024-02-25 10:21:26 +01:00
Sefik Ilkin Serengil
644c6b4013 avoid duplicate detections and loadings 2024-02-24 16:24:15 +00:00
Sefik Ilkin Serengil
0d18a7c2df bug fix in case of load image returns none 2024-02-24 15:35:26 +00:00
Sefik Ilkin Serengil
ab38a1fcf7 throw meaningful error if yolo weights failed while downloading 2024-02-24 15:31:50 +00:00
Andrea Lanfranchi
8472a8b808
Conditional log.info 2024-02-22 14:44:53 +01:00
Andrea Lanfranchi
60c23fbc8e
Remove unnecessary file close 2024-02-22 14:43:01 +01:00
Andrea Lanfranchi
7b0500a17d
Compute toc only if needed 2024-02-22 14:32:16 +01:00
Andrea Lanfranchi
411df327bf
Simplify find data initialization 2024-02-22 14:25:43 +01:00
Andrea Lanfranchi
954ac6e883
Remove trailing white-spaces 2024-02-22 13:01:10 +01:00
Andrea Lanfranchi
5a28a0b0f0
Revert usage of regex 2024-02-22 12:46:22 +01:00
Andrea Lanfranchi
9dc0040a12
Amend typing 2024-02-22 12:41:21 +01:00
Andrea Lanfranchi
63079d3095
Use regex to match file patterns
Also use os.listdir as is clear recursion is not intended
2024-02-22 12:05:41 +01:00
Andrea Lanfranchi
72e5669302
Use path join and amend undetected faces addition repetition 2024-02-21 19:18:01 +01:00
Andrea Lanfranchi
c4b48ce46f
Early discard extracted faces below threshold 2024-02-21 16:28:07 +01:00
Andrea Lanfranchi
051a8d3172
Use return boolean to detect effective capture
Actually an image is always returned but empty
2024-02-21 15:56:47 +01:00
Andrea Lanfranchi
810dfac4fe
Simplify model initialization 2024-02-21 15:18:51 +01:00
Andrea Lanfranchi
adf6328cd9
Look for string in string collection 2024-02-21 15:14:01 +01:00
Andrea Lanfranchi
4713572a24
Amend incompatible assignement warning 2024-02-21 15:12:51 +01:00
Sefik Ilkin Serengil
ab8adf262b
Merge pull request #1026 from serengil/feat-task-1602-loading-image-from-web-bug
load img from web sorted
2024-02-16 17:59:17 +00:00
Sefik Ilkin Serengil
91a220335a exception handling for bulk representations 2024-02-16 17:43:30 +00:00
Sefik Ilkin Serengil
91f5b448ee remove problematic one 2024-02-16 17:39:24 +00:00
Sefik Ilkin Serengil
16c72cd0f6 align first detect second 2024-02-16 17:20:15 +00:00
Sefik Ilkin Serengil
e529fa5c26 make detection, alignment and expanding in common place 2024-02-16 15:31:50 +00:00
Sefik Ilkin Serengil
d16f0765fa load img from web sorted 2024-02-16 13:41:46 +00:00
Sefik Ilkin Serengil
5230943de3 adding package version into package 2024-02-10 18:18:02 +00:00
Sefik Ilkin Serengil
c9569cd540 region bug for dlib sorted 2024-02-10 18:17:28 +00:00
Sefik Ilkin Serengil
626aa2e385 adding some docstrings 2024-02-10 18:17:11 +00:00
Sefik Ilkin Serengil
c65274290b restore version approach 2024-02-04 17:02:31 +00:00
Sefik Ilkin Serengil
36655026cd do not print deepface version error 2024-02-04 13:19:26 +00:00
Sefik Ilkin Serengil
a7231ab0cb adding init py to folders 2024-02-04 12:47:53 +00:00
Sefik Ilkin Serengil
86d424b8ac put api in deepface folder to make it testable 2024-02-04 12:45:27 +00:00
Sefik Ilkin Serengil
08131731c5 retuning regions as numpy int 32 was causing error in api 2024-02-04 10:36:00 +00:00
Sefik Ilkin Serengil
8497682171 open issues 2024-02-02 17:33:31 +00:00
Sefik Ilkin Serengil
96d29ab069 adding expand percentage argument for detection 2024-01-31 23:43:30 +00:00
Sefik Ilkin Serengil
9494d47e31 restoration of deprecated detect face function 2024-01-31 19:51:58 +00:00
Sefik Ilkin Serengil
cefb8013bb making distance calculation functions camel case 2024-01-31 19:44:51 +00:00
Sefik Ilkin Serengil
4800aa3e8c input shape is now retrieved from models 2024-01-31 19:37:43 +00:00
Sefik Ilkin Serengil
95c55c0401 moving logics under modules
- all extract_faces logic moved to detection module
- load image related logics moved to preprocessing module
2024-01-31 19:12:16 +00:00
Sefik Ilkin Serengil
9fbb229b97 Sorting bug for RGB restoration 2024-01-31 19:10:51 +00:00
Sefik Ilkin Serengil
35025cdf6e
Merge pull request #978 from serengil/feat-task-2701-return-obj-for-detectors
pretty return type for detect faces
2024-01-27 19:30:28 +00:00
Sefik Ilkin Serengil
e5f0da3fd7 pretty return type for detect faces 2024-01-27 19:20:03 +00:00
MattPChoy
f0f9a792ec Prevent no attribute error when using Mtcnn 2024-01-27 22:02:10 +10:00
Sefik Ilkin Serengil
36665a9e96 january 26 improvements 2024-01-26 17:52:55 +00:00
Sefik Ilkin Serengil
5ffa7bfb95 vgg normalization layer bug for gpu users 2024-01-23 22:29:28 +00:00
Sefik Ilkin Serengil
3265be2d3a whitespace problem 2024-01-23 20:10:07 +00:00
Sefik Ilkin Serengil
d429d71c93 examples explained in docstrings 2024-01-23 20:02:17 +00:00
Sefik Ilkin Serengil
829871f1fe cosmetic changes in docstrings of interface functions 2024-01-21 22:50:37 +00:00
Sefik Ilkin Serengil
0b3151fa26 incompatible bug for deepid in representation 2024-01-21 18:23:47 +00:00
Sefik Ilkin Serengil
359162aa5e details about return types 2024-01-21 18:10:21 +00:00
Sefik Ilkin Serengil
c9a7db8815 FaceNet to Facenet 2024-01-21 14:01:11 +00:00
Sefik Ilkin Serengil
0d6aa10b82 cosmetic changes about interfaces 2024-01-21 13:52:31 +00:00
Sefik Ilkin Serengil
2e282db74b common function to get tf major version 2024-01-20 20:59:05 +00:00
Sefik Ilkin Serengil
585bdae6b2 DeepFace.build_model should return any 2024-01-20 20:44:00 +00:00
Sefik Ilkin Serengil
a50ff7e35a deprecated module retired 2024-01-20 19:27:08 +00:00
Sefik Ilkin Serengil
25a3061ae9 deprecated dependency retired 2024-01-20 19:26:01 +00:00
Sefik Ilkin Serengil
0fd77e1c99 facial recognition, detector and demography models are now using interface 2024-01-20 19:23:28 +00:00
Sefik Ilkin Serengil
27d15d2413 some improvements
- update datastore
- raise meaningful errors while importing optional dependencies
- handle unexpected zero dimension issue in alignment
2024-01-16 22:31:12 +00:00
Sefik Ilkin Serengil
7f719b87bd interface functions moved to modules 2024-01-13 22:21:41 +00:00
Sefik Ilkin Serengil
0eb1515e11 avoid dimension imcompability error
created pickle may have 2622 dimensional vectors but VGG-Face is not creating
4096 dimensional vectors. If they are mismatch, then raise a meaningful error
2024-01-08 16:59:20 +00:00
Sefik Ilkin Serengil
05013e550c new threshold values for vgg-face
tuned on lfw dataset
2024-01-08 16:57:41 +00:00
Sefik Ilkin Serengil
19f057a712 new structure for vgg-face model 2024-01-08 16:57:07 +00:00
Sefik Ilkin Serengil
f23ab85fd1 using type hinting for backend functions 2023-12-24 14:38:14 +00:00
Sefik Ilkin Serengil
27f0619e9d load img from url was loading rgb instead of bgr 2023-12-24 13:12:35 +00:00
Sefik Ilkin Serengil
308ab61b41 modular tests 2023-12-24 13:06:07 +00:00
Sefik Ilkin Serengil
bd6393daa0
Merge pull request #863 from bhky/master
Simplify face alignment procedure
2023-12-21 09:52:37 +00:00
Sefik Ilkin Serengil
0c0dfb8090
Update DeepFace.py
model type check
2023-12-18 12:06:49 +00:00
Sefik Ilkin Serengil
19f7cf620d
Update DeepFace.py
build_model return type
2023-12-18 11:56:56 +00:00
Sefik Ilkin Serengil
e91e48bd6c
Update DeepFace.py 2023-12-18 09:35:37 +00:00
Sefik Ilkin Serengil
e4a43fd86b
Update DeepFace.py
functional may come from keras or tf.keras. hard to manage in typing. used any for build_model.
2023-12-18 09:31:54 +00:00
Sefik Ilkin Serengil
8a5f2875fc
Merge pull request #922 from RazaProdigy/typehints
Adding Type Hints for Enhanced Code Clarity and Maintenance
2023-12-18 09:30:32 +00:00
RazaProdigy
aa96a76cb2 Removed typo from code 2023-12-18 13:25:36 +04:00
RazaProdigy
90a0282ed9 Refine type hints as per PR review 2023-12-18 13:18:57 +04:00
RazaProdigy
cfff58792a Added type hints 2023-12-17 13:59:32 +04:00
Sefik Ilkin Serengil
bd0a3ec17e yunet alignment issue sorted 2023-12-16 20:11:23 +00:00
xuwei
7b1451ac50 fix(represent): change somthing according to comments and adapt pylint 2023-12-14 12:29:35 +00:00
xuwei
8c6dad0147 fix(represent): 1. support represent user given image 2. img_path support pathlib.Path 2023-12-14 06:57:49 +00:00
Sefik Ilkin Serengil
2cc5f39853 some improvements
- loading image from url was loading as rgb instead of bar
- we are returning image name in load image function to throw
meaningful message for no face found case
- we started to use imread again, and throw exception if file
name contains non English character
- get rid of the duplicate if statement in extract faces
2023-12-08 22:18:52 +00:00
Sefik Ilkin Serengil
76a6bce824 some improvements
- pickle file is out dated, then user informed with meaningful message.
- response df of find, we were just showing source image's region,
we just added the target image's region
2023-12-08 22:15:34 +00:00
Sefik Ilkin Serengil
adfb29dfc7 raise legible error message for optional import 2023-12-08 22:13:20 +00:00
Sefik Ilkin Serengil
fceeccb14b timestamp added for log messages 2023-12-08 22:12:52 +00:00
Sefik Ilkin Serengil
fff6481431 raising legible error message
some libraries are optional. if an user tries to import these
before installation, then not legible message was gotten.
now, we will throw legible message.
2023-12-08 22:12:25 +00:00
Sefik Ilkin Serengil
63b397aa8e review comments resolved 2023-12-06 21:07:31 +00:00
Sefik Ilkin Serengil
125895398e modules added into logger 2023-12-06 20:55:00 +00:00
Sefik Ilkin Serengil
5464ac511c logger instead of print 2023-12-06 19:32:38 +00:00
Sefik Ilkin Serengil
bb2b28e1d9
Update DeepFace.py 2023-11-22 12:02:21 +00:00
Sefik Ilkin Serengil
37c94467f3
Update DeepFace.py 2023-11-22 11:57:55 +00:00
Sefik Ilkin Serengil
f2068df6a8
Update DeepFace.py 2023-11-22 11:49:55 +00:00
Nicolò Di Domenico
c268e396a8
fix missing normalization when skipping face detection 2023-11-15 10:47:24 +01:00
Sefik Ilkin Serengil
8eb8d96a54 old version of model.predict stored 2023-10-22 19:46:02 +01:00
Sefik Ilkin Serengil
68eaafc614
Merge pull request #857 from Villyli/master
change model.predict() to model(), notably improve memory leak (issue #697)
2023-10-22 19:41:11 +01:00
haddyadnan
1bcc065a11 refactor try block 2023-10-19 12:55:58 -07:00
haddyadnan
cca3422b54 Add error msg for failed import 2023-10-19 12:40:36 -07:00
haddyadnan
c53366b75a order result in descending order of probability 2023-10-19 12:09:41 -07:00
haddyadnan
d037d2a2d7 Merge branch 'master' into facenet_pytorch
merge master
2023-10-19 11:32:54 -07:00
haddyadnan
f07be2226f include faster mtcnn implementation 2023-10-19 11:32:39 -07:00
adnanHaddy
0a389deaf1 Include detected face confidence 2023-10-18 12:27:54 -07:00
Bosco Yung
dc75d76813 Simply face alignment procedure 2023-10-18 17:17:34 +09:00
lilw5
1d57b2f7f3 change model.predict() to model(), improve mem leak notably 2023-10-10 11:16:44 +08:00
maxpopovych
50105b21e6 opencvzoo branch changed from master to main, fix applied to sface remote repo 2023-09-08 08:57:12 +02:00
Awshr
1cec1d8f4a find method documentation modified 2023-07-27 09:21:02 +03:00
Anthr@X
9bf9542273 fix linter errors 2023-07-24 15:57:48 +10:00
Anthr@X
fc09756cfc
Merge branch 'master' into fixyunet 2023-07-14 17:16:38 +10:00
Anthr@X
3e7f6dfb58 fix comment length 2023-07-14 14:57:20 +10:00
Anthr@X
a456b43d57 fix line too long 2023-07-14 14:34:05 +10:00
Anthr@X
b97f74e186 fix linting and add comments 2023-07-13 22:13:32 +10:00
Anthr@X
498658d084 Fix yunet return negative coordinates 2023-07-13 11:30:00 +10:00
Sefik Ilkin Serengil
8bb52c0317 some linting comments after yunet implementation 2023-07-09 12:41:31 +01:00
Anthr@X
b52d1e8015 restore original size and move coordinates accordingly 2023-07-09 01:05:17 +10:00
Anthr@X
34d1f1250a return empty list if detected no face. 2023-07-09 00:23:31 +10:00
Anthr@X
a90bf8e1c0 Fix hard coded file name in message 2023-07-08 23:29:28 +10:00