LipNet How can ı start it step-by-step

Yes,question is on the title.Thanks

Mar 28 '19 07:03 HamdiKaptan

Maybe you can start with inferencing with the weights provided?

Mar 29 '19 08:03 DerekChia

Not like that. Which folder ı have to run first

Apr 01 '19 06:04 HamdiKaptan

You will need to:

set up the environment by running pip install -e . in the folder, then
download the GRID dataset, and extract frames with lip actions using scripts/extract_mouth_batch.py
then lastly run python training/overlapped_speakers/train.py s1 to get the training started for person s1 and so on

Apr 01 '19 07:04 DerekChia

Thats what ı really need.

Apr 01 '19 07:04 HamdiKaptan

which one dataset ı have to download ?

Apr 01 '19 07:04 HamdiKaptan

This is using the GRID dataset. Please read the README.md.

This model uses GRID corpus (http://spandh.dcs.shef.ac.uk/gridcorpus/)

Apr 02 '19 16:04 DerekChia

Do i need to make symlinks also? as stated in readme.md

Apr 07 '19 08:04 harshaljain10

@harshaljain10 syslinking doesn't work for me. I'm looking for alternative and might examine the script closer to find out why. Copy and pasting the data should work for now.

Apr 07 '19 15:04 DerekChia

@DerekChia what's mean then lastly run python training/overlapped_speakers/train.py s1 to get the training started for person s1 and so on ?
python training/overlapped_speakers/train.py {s1......s34} ? Does it include s1 s2 s20 s22 ?

Aug 23 '19 01:08 lt573152873

@DerekChia how we load the weights into the model and run the model on a test video

Feb 15 '20 06:02 Ajmalaj11013