Ayush Thakur
  • About Me
  • Authoring
    • DeepFaceDrawing: An Overview
    • Rewriting a Deep Generative Model: An Overview
    • Unsupervised Visual Representation Learning with SwAV
    • In-Domain GAN Inversion for Real Image Editing
    • Metric Learning for Image Search
    • Object Localization with Keras and W&B
    • Image Segmentation Using Keras and W&B
    • Understanding the Effectivity of Ensembles in DL
    • Modern Data Augmentation Techniques for CV
    • Adversarial Latent Autoencoders
    • Towards Deep Generative Modeling with W&B
    • Interpretability in Deep Learning - CAM and GradCAM
    • Introduction to image inpainting with deep learning
    • Simple Ways to Tackle Class Imbalance
    • Debugging Neural Networks with PyTorch
    • Generating Digital Painting Lighting Effects
    • Multi Task Learning with W&B
    • Translate American Sign Language Using CNN
    • Converting FC Layers to Conv Layers
  • Projects
    • Smart Traffic Management Using Reinforcement Learning
    • Sign Language Translator
Powered by GitBook
On this page
  • ​ Read the full article here.
  • ​😼 Check out the GitHub repo here.

Was this helpful?

  1. Authoring

Unsupervised Visual Representation Learning with SwAV

PreviousRewriting a Deep Generative Model: An OverviewNextIn-Domain GAN Inversion for Real Image Editing

Last updated 4 years ago

Was this helpful?

Unsupervised visual representation learning is progressing at an exceptionally fast pace. Most of the modern training frameworks (SimCLR[1], BYOL[2], MoCo (V2)[3]) in this area make use of a self-supervised model pre-trained with some contrastive learning objective. Saying these frameworks perform great w.r.t supervised model pre-training would be an understatement, as evident from the figure below -

Figure 1: Top-1 accuracy of linear classifiers trained with the frozen features of different self-supervised methods w.r.t the fully supervised methods (Source: SwAV [4]).

Moreover, when the features learned using these different self-supervised methods are fine-tuned with as little as 1% and 10% of labeled training data show tremendous performance -

Figure 2: Performance of different semi-supervised and self-supervised frameworks on fine-tuning with very little labeled data (Source: SwAV [4]).

From the above two figures, it is clear that SwAV is currently prevailing in the results (SwAV was published in July 2020), and is currently the SoTA in self-supervised learning for visual recognition. This report will discuss the novel parts that make SwAV such a powerful self-supervised method along with short code walkthroughs.

We expect that you are already familiar with how self-supervised learning works at a high level. If not, this blog post by Jeremy Howard can help you get started.

​😼 Check out the GitHub repo here.

​ Read the full article here.

🔥