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).