[Help wanted] MoCo: Momentum Contrast for Unsupervised Visual Representation Learning

[jaeyounkim]

Help wanted: Research paper code and models

Please help us to implement the following research paper using TensorFlow 2.

If you want to contribute, please leave a comment to express your interest.

Awesome external contributors will be nominated for Google Open Source Peer Bonus.

Research papers

• MoCo: Momentum Contrast for Unsupervised Visual Representation Learning

Status

• Need contribution

Contributor(s)

• Waiting for contributions

Tasks

• Implement a paper in TensorFlow 2
• Reproduce the paper results
• Publish documentation
• Publish pre-trained SavedModel files at TensorFlow Hub

Development environment

• Target
  • Repository: Community directory or Research directory

Requirements

• Follow our contribution guidelines

License

By contributing, you agree that your contributions will be licensed under Apache License 2.0.

[tpapp157]

You can reference my implementation here:
https://github.com/tpapp157/Contrastive_Multiview_Coding-Momentum
It goes beyond just vanilla MoCo and includes some stuff from other papers as well as my own tweaking and improvements.

[ppwwyyxx]

My tensorflow implementation here: https://github.com/ppwwyyxx/moco.tensorflow.
It can reproduce the results in the paper (not just a toy).

[pankajrajput0312]

@jaeyounkim If no one is working on this paper implementation I can start working on it. Can you assign me this issue?

[Dmytruto]

Hi @jaeyounkim, I would like to implement this paper in tf2. Can you assign me to this issue?

[blackburn07x]

@jaeyounkim I would like to work on this paper if no one is working on this. Can you assign this to me?