# Semi-Supervised Semantic Matching

## TL;DR

This paper introduces a semi-supervised learning framework for semantic matching that leverages cyclic consistency constraints on unlabeled image pairs, achieving state-of-the-art results.

## Contribution

It proposes a novel semi-supervised approach with cyclic consistency for semantic matching, addressing the lack of large labeled datasets.

## Key findings

- Achieves state-of-the-art performance on benchmark datasets.
- Effectively utilizes unlabeled data through cyclic consistency.
- Outperforms existing unsupervised and supervised methods.

## Abstract

Convolutional neural networks (CNNs) have been successfully applied to solve the problem of correspondence estimation between semantically related images. Due to non-availability of large training datasets, existing methods resort to self-supervised or unsupervised training paradigm. In this paper we propose a semi-supervised learning framework that imposes cyclic consistency constraint on unlabeled image pairs. Together with the supervised loss the proposed model achieves state-of-the-art on a benchmark semantic matching dataset.