Chainization

October 21, 2022 · View on GitHub

Official pytorch implementation of ECCV 2022 paper, "Order Learning Using Partially Ordered Data via Chainization."

Dependencies

Python 3.8
Pytorch 1.7.1

Datasets

MORPH II

For MORPH II experiments, we follow the same fold settings in this OL repo.

Adience

For Adience experiments, we follow the official splits.

Quick Start : Train Model on Random Edge Cases

You can adjust supervision ratio by changing 'info_ratio' in the parse_option function.

for Adience dataset


    $ python train_chainize_adience.py

for MORPH II dataset


    $ python train_chainize_morph.py

Referecences

FixMatch
POE