can't understand annotation

celebv-text / CelebV-Text

(CVPR 2023) CelebV-Text: A Large-Scale Facial Text-Video Dataset

https://celebv-text.github.io/

388 stars 33 forks source link

can't understand annotation #14

Closed akk-123 closed 9 months ago

akk-123 commented 11 months ago

@zhuhaozh @celebv-text your annotaion file is so confusing，pkl includue act and app like this

[['talk', ['0:0:58', '0:1:16', 18, 0]],
  ['smile', ['0:0:58', '0:1:16', 18, 0]]],

{'scar': [''],
  'mole': [''],
  'freckle': [''],
  'dimple': [''],
  'oneeyed': [''],
  'lighting': [['frontal', ['0:0:58', '0:1:16', 18]]]}

how to get annotation text like paper?

Changezi001 commented 9 months ago

The text descriptions are given in emotion and action files:

emotion: https://drive.google.com/file/d/18V-72aa36r3ZTmu2NHZIMM8aW-0PdKC-/view

action: https://drive.google.com/file/d/1N6Xf2GGstyj--taKnRi5gBd-pXAQX3cI/view

celebv-text commented 9 months ago

Hi @akk-123 , plz see https://github.com/celebv-text/CelebV-Text/issues/16. If you have any questions, plz let me know.