search

serengil / deepface

A Lightweight Face Recognition and Facial Attribute Analysis (Age, Gender, Emotion and Race) Library for Python
https://www.youtube.com/watch?v=WnUVYQP4h44&list=PLsS_1RYmYQQFdWqxQggXHynP1rqaYXv_E&index=1
MIT License
14.34k stars 2.2k forks source link

Model / Detector #818

Closed RafaelFerraria closed 1 year ago

RafaelFerraria commented 1 year ago

I don't know if this is a good place to ask this question. Can anyone tell me/list which is the best model and which is the best detector both in terms of time and accuracy? It's just that when I look and do a search on each of these detectors I get quite confused (opencv, retinaface, mtcnn, ssd,dlib or mediapipe) and then when I go to the models there I am completely lost (VGG-Face, Facenet, Facenet512, OpenFace, DeepFace, DeepID, Dlib,ArcFace, SFace). One more question which metric should I use and what distinguishes them from each other (metrics = ["cosine", "euclidean", "euclidean_l2"])?

serengil commented 1 year ago

Use facenet & mtcnn pair.

Retinaface offers better accuracy but it is very slow. Opencv is yhe fastest but its accuracy is the lowest. Mtcnn offers both fast and accurate results.

serengil commented 1 year ago

Distance metrics offer same results but my favorite is euclidean.