How can I compute the landmarks using my own dataset?

[jingtian-feng]

hi,when you mention about how to prepare dataset,you note that download your pre-computed landmarks.
but when i want to use my own dataset to train ,how can i get the landmarks?
thank you

[mpc001]

Hi, the landmark can be extracted via face tracker. You could check dlib (#1), or face alignment library (#10). We also provide a RetinaFace tracker but it's at another repository.

[seenaimul]

@jingtian-feng did you manage to get it done?
Also how can we do word/sentence level prediction for the video only model? @mpc001

[jingtian-feng]

@jingtian-feng did you manage to get it done? Also how can we do word/sentence level prediction for the video only model? @mpc001

yes，I did , for prediction ,you can refer (#10)

[jingtian-feng]

@jingtian-feng did you manage to get it done? Also how can we do word/sentence level prediction for the video only model? @mpc001

I am glad to communicate with you , for more chat ,you can leave you email or other contact details .
because I am also a beginner

[seenaimul]

@jingtian-feng please connect me on LinkedIn https://www.linkedin.com/in/md-naimul-hasan-5a4259168/
email: [email protected] Looking forward to hearing from you

[seenaimul]

@jingtian-feng can you please share your file for the prediction too? And any additional information to set it up.

can we only check word/sentence prediction for video only model?