Self-Supervised Learning Based Handwriting Verification

TL;DR

This paper explores self-supervised learning methods for handwriting verification, demonstrating that generative and contrastive SSL approaches can outperform supervised methods, especially with limited labeled data.

Contribution

It introduces SSL-HV, applying generative and contrastive SSL techniques to handwriting verification, showing improved accuracy over supervised baselines.

Findings

VAE-based approach achieves 76.3% accuracy.

VICReg fine-tuning achieves 78% accuracy.

SSL methods outperform supervised baseline with fewer labels.

Abstract

We present SSL-HV: Self-Supervised Learning approaches applied to the task of Handwriting Verification. This task involves determining whether a given pair of handwritten images originate from the same or different writer distribution. We have compared the performance of multiple generative, contrastive SSL approaches against handcrafted feature extractors and supervised learning on CEDAR AND dataset. We show that ResNet based Variational Auto-Encoder (VAE) outperforms other generative approaches achieving 76.3% accuracy, while ResNet-18 fine-tuned using Variance-Invariance-Covariance Regularization (VICReg) outperforms other contrastive approaches achieving 78% accuracy. Using a pre-trained VAE and VICReg for the downstream task of writer verification we observed a relative improvement in accuracy of 6.7% and 9% over ResNet-18 supervised baseline with 10% writer labels.