Machine learning models

This page lists the provided machine learning models that are pre-trained and ready to use for inference. The weights of the models are located by default on data/models/ and can be downloaded from the GitHub releases.

The output of the models is composed of Toolbox Structures.

Face analysis

Name	Description	Output
age_gender	Predict the age and gender of a face image	(Gender, float)
emotions_hse	Recognize 7 different emotions from a face image	Emotion
face_recognition_facenet	Perform face recognition tasks	(str, float) \| np.ndarray

Object detection

Name	Description	Output
face_detector_retinaface	Detect faces in images	BoundingBox
face_detector_ultraface	Detect faces in images	BoundingBox

Instance segmentation

Name	Description	Output
detectron2	Perform instance segmentation of 80 different objects	SegmentationMask

Keypoints

Name	Description	Output
detectron2	Predict the position of 17 body key points	COCOKeypoints