Benchmarking Detection Transfer Learning with Vision Transformers

TL;DR

This paper benchmarks Vision Transformer models for object detection transfer learning, overcoming training challenges, and finds that masking-based self-supervised pre-training improves detection accuracy on COCO, especially for larger models.

Contribution

It introduces training techniques enabling ViT models as backbones in detection tasks and compares various initializations, highlighting the benefits of masking-based self-supervised learning.

Findings

Masking-based self-supervised pre-training improves detection accuracy.

Improvements increase with larger model sizes.

The methods enable effective use of ViT in detection benchmarks.

Abstract

Object detection is a central downstream task used to test if pre-trained network parameters confer benefits, such as improved accuracy or training speed. The complexity of object detection methods can make this benchmarking non-trivial when new architectures, such as Vision Transformer (ViT) models, arrive. These difficulties (e.g., architectural incompatibility, slow training, high memory consumption, unknown training formulae, etc.) have prevented recent studies from benchmarking detection transfer learning with standard ViT models. In this paper, we present training techniques that overcome these challenges, enabling the use of standard ViT models as the backbone of Mask R-CNN. These tools facilitate the primary goal of our study: we compare five ViT initializations, including recent state-of-the-art self-supervised learning methods, supervised initialization, and a strong random initialization baseline. Our results show that recent masking-based unsupervised learning methods may, for the first time, provide convincing transfer learning improvements on COCO, increasing box AP up to 4% (absolute) over supervised and prior self-supervised pre-training methods. Moreover, these masking-based initializations scale better, with the improvement growing as model size increases.