8000 GitHub - matkir/yousaidthat-tf
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

matkir/yousaidthat-tf

BranchesTags
 
 

Repository files navigation

code07-server009

  • 使用所有pre-trained参数

code06-server009

  • 修改一些细节

code05-server009

  • 核对超参数
  • 修改build_data.py: python build_data.py --device --func --dataset --checkpoint

code04-server009

使用v201中pre trained identity encoder参数

  1. python build_data.py --device mac --func txt: 生成所有MP4路径的TXT
  2. python build_data.py --device mac --func data: 1个MP4可以生成5个NPZ (face + mfcc + identity).
  3. python build_data.py --device mac --func tfrecords: NPZ --> TFRECORDS
  4. python speech2vid_train_finetune.py --tfrecords /path/to/tfrecords

code03

  1. python build_data.py --device mac --func txt: 生成所有MP4路径的TXT
  2. python build_data.py --device mac --func data: 1个MP4可以生成5个NPZ (face + mfcc + identity).
  3. python build_data.py --device mac --func tfrecords: NPZ --> TFRECORDS
  4. python speech2vid_train.py --tfrecords /path/to/tfrecords

code02

  1. python build_data.py --device mac --func txt: 生成所有MP4路径的TXT
  2. python build_data.py --device mac --func data: 1个MP4可以生成5个NPZ (face + mfcc + identity).
  3. python build_data.py --device mac --func tfrecords: NPZ --> TFRECORDS

code01

在mac上进行测试：

  1. prepare_data.py: MP4 --> NPZ face + mfcc + identity
  2. build_data.py: NPZ --> TFRECORDS
  3. speech2vid_train.py: TRAIN

About

No description, website, or topics provided.

Resources

Readme
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%
0