yinguobing / cnn-facial-landmark

Training code for facial landmark detection based on deep convolutional neural network.
MIT License
623 stars 183 forks source link

new Dataset #108

Closed MacBook-Pro-gala closed 3 years ago

MacBook-Pro-gala commented 3 years ago

How to train this model with the new dataset 300WV?

yinguobing commented 3 years ago

Is that a public dataset? Could you post the link here?

MacBook-Pro-gala commented 3 years ago

Is that a public dataset? Could you post the link here? https://ibug.doc.ic.ac.uk/resources/300-VW/

yinguobing commented 3 years ago

You mean 300VW!

That dataset has the exactly same mark format as 300W. Convert the videos into frame images first. The training process is identical to that of the 300W dataset.