Releases: serengil/deepface
v0.0.99
Contributors
Assets 2
v0.0.98
What's Changed
- using digital signatures for pickle creation and loading by @serengil in #1570
- pgvector backend added by @serengil in #1572
- raise error while attempting signed datastore without credentials by @serengil in #1576
- add pinecone as backend db by @serengil in #1577
- manage database inventory dynamically by @serengil in #1578
- temp folder decided by the tempfile module now by @serengil in #1580
- spoofing module class bug sorted by @serengil in #1581
Full Changelog: v0.0.97...v0.0.98
Contributors
Assets 2
v0.0.97
What's Changed
- make return types raw python for landmarks by @serengil in #1532
- mypy initialized by @serengil in #1533
- batch extract faces by @serengil in #1534
- throw custom errors by @serengil in #1535
- install dependencies once by @serengil in #1536
- more unit tests added for recent batch extract faces feature by @serengil in #1538
- add similarity search support to find function by @serengil in #1540
- Stateless Search Functionality Added by @serengil in #1544
- sorting broken tests by @serengil in #1545
- load models on app start up to respond faster in 1st request by @serengil in #1548
- in case of no embedding, we were adding distance but skipped confidence by @serengil in #1551
- mongo support added by @serengil in #1550
- weaviate added by @serengil in #1554
- db con can be set by env vars, also provide confidence and detected face area in search's result by @serengil in #1555
- return img_name for faiss searches by @serengil in #1557
- bearer token support added by @serengil in #1558
- broken yolo links updated by @serengil in #1560
- more tests added for authorization and connection string for api by @serengil in #1561
- simplify auth service by @serengil in #1563
- multi model support with different dim size added for weaviate by @serengil in #1565
- neo4j backend added by @serengil in #1567
Full Changelog: v0.0.96...v0.0.97
Contributors
Assets 2
v0.0.96
What's Changed
- store configurations in seperate folder not in main code by @serengil in #1503
- replace training.Model with Model in ArcFace by @niteshsahujhansi in #1507
- Expose AVAILABLE_MODELS from build_model. by @timohl in #1514
- Yolo 12 models added, eye detection is done in global point if not available by @serengil in #1518
- output normalization functionalities added to represent by @serengil in #1526
- encrypt module added by @serengil in #1527
New Contributors
- @niteshsahujhansi made their first contribution in #1507
- @timohl made their first contribution in #1514
Full Changelog: v0.0.95...v0.0.96
Contributors
Assets 2
v0.0.95
What's Changed
- fbdeepface's confidence metric added by @serengil in #1498
- Fix FastMtCnn detection error by @nico-hahn in #1499
- Revert flask and werkzeug dependency changes by @pcbowers in #1502
New Contributors
- @nico-hahn made their first contribution in #1499
- @pcbowers made their first contribution in #1502
Full Changelog: v0.0.94...v0.0.95
Contributors
Assets 2
v0.0.94
What's Changed
- Feat task 1708 stream overlay bug for crowded photo by @serengil in #1311
- Dataclass by @kremnik in #1317
- Represent optimizations by @Circuit8 in #1319
- Optimize ssd detector by @kremnik in #1321
- weight downloads moved to a common place by @serengil in #1325
- cast confidence to float by @serengil in #1326
- load weights is done in a common function by @serengil in #1327
- unit test for broken weight file by @serengil in #1329
- Tiny fix: changed index to label in ssd detector by @kremnik in #1330
- Improvement: check image exists first by @kremnik in #1334
- unit tests added for weight utils by @serengil in #1337
- Normalized paths in test_commons.py to make tests platform-independent by @kremnik in #1350
- Recognition find batched by @kremnik in #1351
- Feat task 0510 developments by @serengil in #1357
- all models can be downloaded with one shot by @serengil in #1358
- single and batch distance functions are stored in verify module by @serengil in #1359
- Configurable mediapipe params by @tyasnk in #1362
- min detection confidence score retrieved from env vars by @serengil in #1364
- improve Dockerfile to use entrypoint in order to get better log when … by @seavphing in #1369
- Update requirements.txt by @cobycloud in #1370
- Update Dockerfile by @mone855 in #1374
- Update image_utils.py by @sachadee in #1377
- Feat task 0911 file input for api by @serengil in #1383
- more unit tests added by @serengil in #1386
- [FEATURE]: adding yolov11 into face detection portfolio by @nriviera in #1397
- Fix: code comment correction by @dohyeonYoon in #1402
- Fix: model code comment correction by @dohyeonYoon in #1405
- Update CenterFace threshold by @mehran66 in #1407
- Improve the face alignment performance in detect_faces() by @huulockt in #1409
- fastmtcnn's coordinates casted to int by @serengil in #1411
- video path is enabled in stream by @Mehrab-Shahbazi in #1415
- pickle objects are now dumped with pickle.HIGHEST_PROTOCOL by @PyWoody in #1418
- fix small spelling error :)) by @gustavn19 in #1421
- Optimizations for deepface.recognition.find, Optimization and New Iterator Functionality in image_utils by @PyWoody in #1420
- load_image now accepts file objects that support being read by @PyWoody in #1423
- Feat/batch predict age and gender by @NatLee in #1396
- Batching on
.representto improve performance and utilize GPU in full by @galthran-wq in #1433 - post batch changes by @serengil in #1438
- Update representation.py by @dakotah-jones in #1427
- post batch changes by @serengil in #1440
- allow 0 (as integers) when using pre-calculated embeddings in verify by @jabiinfante in #1446
- added buffalo_l model to deepface by @Raghucharan16 in #1439
- Updated Readme for buffalo_l by @Raghucharan16 in #1449
- Changing of thresholds for buffalo_l by @Raghucharan16 in #1450
- Update VGGFace.py by @timselier in #1456
- feature: Add Angular Distance as a Distance Metric by @haddyadnan in #1459
- fix: typo in error message for missing pytorch by @CatBraaain in #1463
- feeding clean images to ml models by @serengil in #1470
- Update README.md by @arsaIan in #1472
- Fix AttributeError in streaming by @catherinetcai in #1475
- Diagram visualization of the deepface codebase by @ivanmilevtues in #1479
- sanitized facial landmarks, and test for facial landmark sanitizaion by @RUTUPARNk in #1484
- is_valid_landmark in test_landmark_sanitization file by @RUTUPARNk in #1485
- patched mockes' types specified by @serengil in #1487
- Pass test suite with python 3.8 by @ddaanet in #1492
- distance to confidence calculation added by @serengil in #1497
New Contributors
- @Circuit8 made their first contribution in #1319
- @tyasnk made their first contribution in #1362
- @seavphing made their first contribution in #1369
- @cobycloud made their first contribution in #1370
- @mone855 made their first contribution in #1374
- @sachadee made their first contribution in #1377
- @nriviera made their first contribution in #1397
- @dohyeonYoon made their first contribution in #1402
- @mehran66 made their first contribution in #1407
- @huulockt made their first contribution in #1409
- @Mehrab-Shahbazi made their first contribution in #1415
- @PyWoody made their first contribution in #1418
- @gustavn19 made their first contribution in #1421
- @NatLee made their first contribution in #1396
- @galthran-wq made their first contribution in #1433
- @dakotah-jones made their first contribution in #1427
- @jabiinfante made their first contribution in #1446
- @Raghucharan16 made their first contribution in #1439
- @timselier made their first contribution in #1456
- @CatBraaain made their first contribution in #1463
- @arsaIan made their first contribution in #1472
- @catherinetcai made their first contribution in #1475
- @ivanmilevtues made their first contribution in #1479
- @RUTUPARNk made their first contribution in #1484
- @ddaanet made their first contribution in #1492
Full Changelog: v0.0.93...v0.0.94
Contributors
Assets 2
v0.0.93
🚀 DeepFace v0.093 Release Notes
We’re excited to announce the latest release of DeepFace, packed with enhancements, bug fixes, and new features. Here’s what’s new in v0.093:
🌐 New Features
- CORS Enabled for All Routes: Now, CORS is enabled across all routes, enhancing cross-origin requests. (#1257)
- Face Extraction Options: Added an option to extract_faces to return faces in RGB, BGR, or Grayscale formats. More flexibility for your image processing! (#1279)
- Batch Prediction for Demography Models: We've introduced batch prediction support for demography models, speeding up the process. (#1298)
- Max Faces Argument: A new max_faces argument in represent lets you limit the number of faces processed in large images. (#1283)
🔧 Enhancements
- Image Quality Improvement: The resample argument is now set to Image.BICUBIC when rotating faces, improving the quality of low-resolution images. (#1269)
- Global Variable Management: We've centralized the management of global variables for better efficiency and maintainability. (#1296)
- Logger Class Update: The logger class is now fully Singleton, ensuring consistent logging throughout the application. (#1291)
🛠️ Refactorings
- Distance Calculation Optimization: Distance calculation functions are now more efficient, using fewer commands and leveraging numpy. (#1299)
- Detection Module Revamp: The DetectorWrapper is retired; detection is now handled directly within the detection module. (#1302)
- Model Structure Overhaul: Facial recognition, face detector, demography, and spoofing models are now organized into dedicated folders under the models directory. (#1303)
- File System Command Refactor: We’ve streamlined file system commands for better performance. (#1305)
- Facial Database Optimization: We now use set instead of list when creating the facial database, improving performance. (#1306)
- Verification Module Refactor: The verification module has been refactored to improve readability and efficiency. Previously, the same code block was run twice for an image pair; now, the logic has been moved to a subfunction, which is called separately for each image. This change simplifies the code and enhances maintainability. (#1309)
🐛 Bug Fixes
- Grayscale Face Bug Fix: Resolved an issue that occurred when returning a face in grayscale. (#1276)
- Facial Area Coordinate Bug: Fixed a bug where facial area coordinates could sometimes fall outside image borders. (#1308)
Thank you for your continued support and contributions to DeepFace! 🎉
Packages
v0.0.92
DeepFace 0.0.92 Release Notes 🎉
New Features & Improvements 🚀
-
Enhanced Find Function 🔍
Added refresh_database argument to the find function. The default value is True, but when set to False, a pickle file is used without checking for differences. (#1235) -
Version Display 📋
Package version is now shown on the API's homepage and in console logs, making it easier to track which version you're using. (#1246) -
Service Port Update 🔄
Service port moved from 5000 to 5005 to avoid conflicts with AirDrop on MacBooks. (#1246) -
Image Alignment Improvement 🖼️
When faces are close to the borders, alignment previously moved them outside the image. A border with black pixels is now added to avoid this issue. (#1247) -
Official Docker Image 🐳
Pushed the official DeepFace image to DockerHub for easier deployment and consistency. (#1249) -
Face Anti-Spoofing 🔒
Added a face anti-spoofing feature to enhance security and prevent fraudulent face inputs. (#1253)
v0.0.91
🎉 DeepFace v0.0.91 Release Notes 🚀
We're excited to introduce some fantastic updates in this release! Here's what's new:
📏 Custom Threshold Support: With PR-1223, we've enhanced the verify method to support an optional threshold parameter. Now, you can tailor the verification process to your needs for stricter or more lenient results.
📚 Enhanced Documentation: Thanks to PR-1224, you'll find the latest publication mentioned in the README. Additionally, experiment results are conveniently shared in the benchmarks folder for your reference.
🖼️ Improved Functionality: We've fixed a bug in the represent function (PR-1225) where face regions were inaccurately returned if detection was skipped. Now, you can expect more reliable results.
📝 License Inclusion: PR-1225 addresses a missing license in the pip package, ensuring compliance and transparency.
🔧 Issue Templates: To streamline communication and issue tracking, we've created issue templates with PR-1203. Now, reporting and addressing concerns is easier than ever before.
Thank you for your continued support and feedback! Download DeepFace v0.0.91 now to enjoy these enhancements and more. Happy coding! 🤖✨
Package
v0.0.90
🚀 DeepFace 0.0.90 Release Notes
👁️ Improved Eye Detection:
-
Now utilizes dlib detector for obtaining the centers of the left and right eyes, enhancing accuracy and reliability (#1138).
-
Eye positioning is now relative to the person instead of the observer, aligning with biological standards (#1140).
💻 Enhanced Performance:
- Fast MtCnn detector now supports GPU acceleration for faster processing (#1145).
🔍 Model Correction:
- FaceNet model now correctly uses InceptionResNetV1, rectifying the mislabeling as V2 in its docstring (#1151).
🔧 Streamlined Functionality:
-
Embeddings extraction function renamed to "forward" and moved to superclass for improved modularity (#1156).
-
Improved file detection by examining content type instead of relying solely on file extension, enhancing security (#1171, #1173, #1174).
-
Separated resize functionality into representation module for better code organization (#1175).
-
Singleton pattern implemented for logger classes to optimize resource usage (#1177).
-
Introduction of CenterFace face detector to DeepFace's toolkit (#1184).
-
Consideration of additional arguments impacting embeddings during facial database creation (#1185).
-
Introduction of File Utils for improved file management (#1186).
🛠️ Bug Fixes:
- Detection post-alignment now validates projected coordinates against base image size to prevent failures due to invalid coordinates (#1191).