CoDo: Contrastive Learning with Downstream Background Invariance for Detection

TL;DR

CoDo introduces a novel self-supervised learning approach focusing on background invariance at the object level, significantly improving transfer performance for object detection tasks compared to traditional instance discrimination methods.

Contribution

The paper proposes a new object-level self-supervised learning method, CoDo, emphasizing background invariance and architecture alignment to enhance object detection performance.

Findings

Achieves strong transfer learning results on MSCOCO

Outperforms traditional instance discrimination methods

Demonstrates effectiveness with ResNet50-FPN backbone

Abstract

The prior self-supervised learning researches mainly select image-level instance discrimination as pretext task. It achieves a fantastic classification performance that is comparable to supervised learning methods. However, with degraded transfer performance on downstream tasks such as object detection. To bridge the performance gap, we propose a novel object-level self-supervised learning method, called Contrastive learning with Downstream background invariance (CoDo). The pretext task is converted to focus on instance location modeling for various backgrounds, especially for downstream datasets. The ability of background invariance is considered vital for object detection. Firstly, a data augmentation strategy is proposed to paste the instances onto background images, and then jitter the bounding box to involve background information. Secondly, we implement architecture alignment between our pretraining network and the mainstream detection pipelines. Thirdly, hierarchical and multi views contrastive learning is designed to improve performance of visual representation learning. Experiments on MSCOCO demonstrate that the proposed CoDo with common backbones, ResNet50-FPN, yields strong transfer learning results for object detection.