search

open-mmlab / mmselfsup

OpenMMLab Self-Supervised Learning Toolbox and Benchmark
https://mmselfsup.readthedocs.io/en/latest/
Apache License 2.0
3.14k stars 429 forks source link

Unlabeled data preparation #758

Open opencomvis opened 1 year ago

opencomvis commented 1 year ago

Given a dataset that contains only 20 percent of annotated data divided into train/valid. How can we prepare the remaining 80% of the dataset that doesn't has labels ?