iteal / wormpose

WormPose: Image synthesis and convolutional networks for pose estimation in C. elegans
https://iteal.github.io/wormpose/
BSD 3-Clause "New" or "Revised" License
46 stars 16 forks source link

Reading a Tierpsy h5 file #21

Closed WeiWenKevinChen closed 1 year ago

WeiWenKevinChen commented 2 years ago

Hi, I am trying to use your WormPose to analyze the video that contains multiple worms. I first used the Tierpsy to extract the individual worms but I am not sure how to input different worms in the same Tierpsy h5 file to the WormPose. Could you let me know how to do it? Thanks!

iteal commented 2 years ago

Hi! I think you need first to decide if the different worms are similar or different enough visually, so that you can train a network from images of each of them separately or together. At https://iteal.github.io/wormpose/tierpsy.html there is some information on how to organize the Tierpsy files to use with wormpose. What we did is one worm per video (two hdf5 files one for images one for features), and several videos to constitute a wormpose dataset.

WeiWenKevinChen commented 2 years ago

I see. Thanks for your reply!

iteal commented 2 years ago

Actually, only one worm per file is supported at the moment, so if there are several worms in the same hdf5 file. only the first one will be used.