Self-Supervised Learning

Learn Visual Representations with Masked Autoencoders

Labeled image datasets are expensive to build, but unlabeled images are often plentiful. A useful pretraining strategy is therefore to create a learning signal from each image itself instead of asking a human to annotate it. A masked autoencoder (MAE) does this by hiding part of an image and training a model to reconstruct the missing content. The reconstruction task is not usually the final product. Its purpose is to make the encoder learn visual representations that can later support tasks such as classification or detection.