search

lippman1125 / caffe_s3fd

caffe for s3fd face detection
Other
32 stars 13 forks source link

how's the inference speed #1

Open lucasjinreal opened 5 years ago

lucasjinreal commented 5 years ago

how about the inference speed of this model?

lippman1125 commented 5 years ago

vgg16+s3fd hard to be realtime on mobile device, if you want a realtime model, you can you mv2+s3fd. https://github.com/lippman1125/S3FD.PyTorch

lucasjinreal commented 5 years ago

@lippman1125 Not mention mobile, I have test s3fd on GTX 1080TI, it needs about 0.2s per image, which is very slow....

Does this normal?

thanks for your pytorch implementation! seems mobilenet plus s3fd may got reasonable speed! How does the accuracy drop compares to vgg backbone through?

jamessmith90 commented 4 years ago

@lippman1125 Can you share the resnet18 model on Google Drive ?

Thanks.