You Don’t Need Domain-Specific Data Augmentations When Scaling Self-Supervised Learning
Keywords: dinov2, self-supervised learning, data-augmentations, scaling law
TL;DR: Self-Supervised Learning (SSL) doesn't need hand-crafted data-augmentations, given enough amount of data. It's possible to train a powerful DINOv2 model with less augmentations than all other modern SSL models.
Abstract: Self-Supervised learning (SSL) with Joint-Embedding Architectures (JEA) has led to outstanding performances. All instantiations of this paradigm were trained using strong and well-established hand-crafted data augmentations, leading to the general belief that they are required for the proper training and performance of such models. On the other hand, generative reconstruction-based models such as BEIT and MAE or Joint-Embedding Predictive Architectures such as I-JEPA have shown strong performance without using data augmentations except masking. In this work, we challenge the importance of invariance and data-augmentation in JEAs at scale. By running a case-study on a recent SSL foundation model -- DINOv2 -- we show that strong image representations can be obtained with JEAs and only cropping without resizing provided the training data is large enough, reaching state-of-the-art results and using the least amount of augmentation in the literature. Through this study, we also discuss the impact of compute constraints on the outcomes of experimental deep learning research, showing that they can lead to very different conclusions.
Primary Area: Machine vision
Submission Number: 10849
Loading