Exploring the Limits of Weakly Supervised Pretraining

Authors:
  • DMDhruv Mahajan
    Dhruv Mahajan
  • RGRoss Girshick
    Ross Girshick
  • VRVignesh Ramanathan
    Vignesh Ramanathan
  • KHKaiming He
    Kaiming He
  • MPManohar Paluri
    Manohar Paluri
  • YLYixuan Li
    Yixuan Li
  • ABAshwin Bharambe
    Ashwin Bharambe
  • LMLaurens van der Maaten
    Laurens van der Maaten
Preprint
Publisher website
Google Scholar
Abstract
State-of-the-art visual perception models for a wide range of tasks rely on supervised pretraining. ImageNet classification is the de facto pretraining task for these models. Yet, ImageNet is now nearly ten years old and is by modern standards "small". Even so, relatively little is known about the behavior of pretraining with datasets that are multiple orders of magnitude larger. The reasons are obvious: such datasets are difficult to collect and annotate. In this paper, we present a unique study of transfer learning with large convolutional networks trained to predict hashtags on billions of social media images. Our experiments demonstrate that training for large-scale hashtag prediction leads to excellent results. We show improvements on several image classification and object detection tasks, and report the highest ImageNet-1k single-crop, top-1 accuracy to date: 85.4% (97.6% top-5). We also perform extensive experiments that provide novel empirical data on the relationship between large-scale pretraining and transfer learning performance.
All Related Versions