Self-Supervised Learning of Pretext-Invariant Representations

TL;DR

PIRL is a self-supervised learning method that learns invariant image representations, significantly improving semantic quality and outperforming supervised pre-training on benchmarks like object detection.

Contribution

We introduce PIRL, a novel self-supervised approach that enforces invariance to transformations, enhancing semantic representation quality beyond existing methods.

Findings

PIRL achieves state-of-the-art results on self-supervised benchmarks.

PIRL outperforms supervised pre-training in object detection tasks.

PIRL improves the semantic invariance of learned representations.

Abstract

The goal of self-supervised learning from images is to construct image representations that are semantically meaningful via pretext tasks that do not require semantic annotations for a large training set of images. Many pretext tasks lead to representations that are covariant with image transformations. We argue that, instead, semantic representations ought to be invariant under such transformations. Specifically, we develop Pretext-Invariant Representation Learning (PIRL, pronounced as "pearl") that learns invariant representations based on pretext tasks. We use PIRL with a commonly used pretext task that involves solving jigsaw puzzles. We find that PIRL substantially improves the semantic quality of the learned image representations. Our approach sets a new state-of-the-art in self-supervised learning from images on several popular benchmarks for self-supervised learning. Despite being unsupervised, PIRL outperforms supervised pre-training in learning image representations for object detection. Altogether, our results demonstrate the potential of self-supervised learning of image representations with good invariance properties.