Document Domain Randomization for Deep Learning Document Layout Extraction

TL;DR

This paper introduces document domain randomization (DDR), a novel approach that trains CNNs on rendered pseudo-documents to effectively transfer to real-world document segmentation tasks, demonstrating robustness across styles and sample sizes.

Contribution

The paper presents the first successful transfer of CNNs trained solely on rendered pseudo-documents to real-world document segmentation, enabling effective learning without real data.

Findings

DDR achieves competitive results on benchmark datasets.

Style mismatch impacts model accuracy more than label noise.

Smaller training samples slightly reduce performance, but high accuracy persists.

Abstract

We present document domain randomization (DDR), the first successful transfer of convolutional neural networks (CNNs) trained only on graphically rendered pseudo-paper pages to real-world document segmentation. DDR renders pseudo-document pages by modeling randomized textual and non-textual contents of interest, with user-defined layout and font styles to support joint learning of fine-grained classes. We demonstrate competitive results using our DDR approach to extract nine document classes from the benchmark CS-150 and papers published in two domains, namely annual meetings of Association for Computational Linguistics (ACL) and IEEE Visualization (VIS). We compare DDR to conditions of style mismatch, fewer or more noisy samples that are more easily obtained in the real world. We show that high-fidelity semantic information is not necessary to label semantic classes but style mismatch between train and test can lower model accuracy. Using smaller training samples had a slightly detrimental effect. Finally, network models still achieved high test accuracy when correct labels are diluted towards confusing labels; this behavior hold across several classes.