Ibot : Image Bert Pre-Training with Online Tokenizer

By J. Zhou et al.

Published on June 10, 2022

Read the original document by opening this link in a new tab.

Table of Contents

1. Introduction
2. Preliminaries
3. iBOT
4. Experiment

Summary

The paper introduces iBOT, a framework for masked image modeling using an online tokenizer. It presents a self-supervised framework that performs masked prediction with an online tokenizer. The online tokenizer is jointly learnable with the masked image modeling objective, eliminating the need for pre-training. Results show significant advancements in image classification and robustness against common corruptions.