Self-supervised Learning with Local Contrastive Loss for Detection and Semantic Segmentation

TL;DR

This paper introduces a self-supervised learning method that uses a local contrastive loss to improve feature consistency for detection and segmentation tasks, achieving state-of-the-art results.

Contribution

The proposed local contrastive loss can be integrated into existing SSL methods to enhance performance on semi-global tasks like detection and segmentation.

Findings

Outperforms state-of-the-art SSL methods by 1.9% on COCO detection

Achieves 1.4% improvement on PASCAL VOC detection

Gains 0.6% on CityScapes segmentation

Abstract

We present a self-supervised learning (SSL) method suitable for semi-global tasks such as object detection and semantic segmentation. We enforce local consistency between self-learned features, representing corresponding image locations of transformed versions of the same image, by minimizing a pixel-level local contrastive (LC) loss during training. LC-loss can be added to existing self-supervised learning methods with minimal overhead. We evaluate our SSL approach on two downstream tasks -- object detection and semantic segmentation, using COCO, PASCAL VOC, and CityScapes datasets. Our method outperforms the existing state-of-the-art SSL approaches by 1.9% on COCO object detection, 1.4% on PASCAL VOC detection, and 0.6% on CityScapes segmentation.