Skip to content
Markerless pose estimation of user-defined features with deep learning for all animals, including humans
Python Jupyter Notebook Shell
Branch: master
Clone or download
Pull request Compare This branch is 31 commits ahead, 7 commits behind AlexEMG:master.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
.github/ISSUE_TEMPLATE
conda-environments
deeplabcut
docs
examples
.gitignore
AUTHORS
CODE_OF_CONDUCT.md
LICENSE
README.md
README.org.md
dlc.py
dlc_tpu.ipynb
reinstall.sh
setup.py

README.md

DeepLabCut for TPU

Markerless pose estimation of user-defined features with deep learning for all animals, including humans

for Google Colaboratory TPU

Original README.md

Modification

  • Change model for tf.data.Dataset and tf.estimator
  • Change training dataset for TPU
    • Fix image size
    • Use tensorflow's image distortion functions
  • Change training dataset for improve performance
    • Training images are randomly rotated
    • Training images are randomly changed brightness and contrast
  • Training steps are visualized for tensorboard
  • For TPU, 'analyze_videos' function are changed use tfrecord converted inputs
  • For TPU, add 'convert_analyze_videos' function for convert to tfrecord input
  • Change 'create_labeled_video' function for alpha labels marking

How to use in Cloaboratory TPU

  1. process to label_frames steps.
  2. open Colaboratory notebook and follow the instructions

How to use in local system

same as original

You can’t perform that action at this time.