deepcam-cn / yolov5-face

YOLO5Face: Why Reinventing a Face Detector (https://arxiv.org/abs/2105.12931) ECCV Workshops 2022)
GNU General Public License v3.0
2.01k stars 491 forks source link

two questions about this excellent project #254

Open JH-Lam opened 1 year ago

JH-Lam commented 1 year ago

hi , thanks for your effort to raise this project. now I have a couple of equestions there:

  1. where is the model yolov5s6? and what does the word 's6' mean?
  2. are the calculation all correct in the table 'pre-trained models'? since I recalculated the yolov5s-face.pt model but it was not same as yours:

yolov5s | FLOPs 5.751 | parmas 7.075 ,

mine (via thop with 640 imgsize):

yolov5s | FLOPs 7.62| parmas 7.06,

thanks again