facebookresearch/swav
A PyTorch implementation of SwAV, a self-supervised learning method for pre-training convolutional neural networks on visual data without annotations.
Not currently ranked — collecting fresh signals.
star history
SwAV learns visual representations by simultaneously clustering data and enforcing consistency between cluster assignments produced for different augmentations of the same image. The method uses a swapped prediction mechanism where it predicts the cluster assignment of one view from the representation of another. It provides pre-trained ResNet-50 models and can be trained with both large and small batches, scaling to unlimited amounts of unlabeled data.
Frequently asked
- What is facebookresearch/swav?
- A PyTorch implementation of SwAV, a self-supervised learning method for pre-training convolutional neural networks on visual data without annotations.
- Is swav open source?
- Yes — facebookresearch/swav is an open-source project tracked on heatdrop.
- What language is swav written in?
- facebookresearch/swav is primarily written in Python.
- How popular is swav?
- facebookresearch/swav has 2.1k stars on GitHub.
- Where can I find swav?
- facebookresearch/swav is on GitHub at https://github.com/facebookresearch/swav.