Deep Neural Networks for Page Stream Segmentation and Classification

Gallo, Ignazio; Noce, Lucia; Zamberletti, Alessandro; Calefati, Alessandro

doi:10.1109/dicta.2016.7797031

Cited by 9 publications

(10 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Gallo et al proposed in their study a hybrid technique using CNN, followed by a Deep Neural Network (DNN) to extract the visual features of the text and classified the documents. The page stream segmentation was done using the document classification in the proposed method [4].…”

Section: Related Workmentioning

confidence: 99%

“…We have traced the evolution (Table 1) of the DSS technologies starting from the stochastic Markov chain model in 2009 [3] through the deep image-based page feature extraction and classification in 2016 [4], rule-based approach in 2017 [5] to a more sophisticated state-of-the-art multi-modal deep learning approach combining text and image features of the document page until 2021 [6]. Although, there is a clear dearth of the overall study observed in the domain of DSS, the recent breakthrough in the domain of DSS by Wiedemann and Heyer [7] shows promising results with Tobacco800 public data set and a proprietary data set.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Multi-Modal Approach to Digital Document Stream Segmentation for Title Insurance Domain

et al. 2022

View full text Add to dashboard Cite

In the twenty-first century, storing and managing digital documents has become commonplace for all corporate and public sectors around the world. Physical documents are scanned in batches and stored in a digital archive as a heterogeneous document stream, referred to as a digital package. To make Robotic Process Automation (RPA) easier, it's necessary to automatically segment the document stream into a subset of independent, coherent multi-page documents by detecting the appropriate document boundary. It's a common requirement of a TI company's Automated Document Management Systems (ADMS), where business operations are automated using RPA and the goal is to extract information from digital documents with minimal user intervention. The current study proposes, evaluates, and compares a multi-modal binary classification network incorporating text and picture aspects of digital document pages to state-of-the-art baseline methodologies. Image and textual features are extracted simultaneously from the input document image by passing them through Visual Geometry Group 16 -Convolutional Neural Network (VGG16-CNN) and pre-trained Bidirectional Encoder Representations from Transformers (Legal-BERT base ) model through transfer learning respectively. Both features are finally fused and passed through a fully connected layer of Multi Layered Perceptron (MLP) to obtain the binary classification of the pages as the First Page (FP) and Other Page (OP). Real-time document image streams from production business process archive were obtained from a reputed Title Insurance (TI) company for the study. The obtained F 1 score of 97.37% and 97.15% are significantly higher than the accuracies of the considered two baseline models and well above the expected Straight Through Pass (STP) threshold defined by the process admin.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A Multi-Modal Approach to Digital Document Stream Segmentation for Title Insurance Domain

et al. 2022

View full text Add to dashboard Cite

show abstract

“…In this case, the system is iterated on the successive pairs of pages, takes as input the current pair of pages, looks at whether the two pages contain indices of similarity or continuity, then, decides whether they represent a rupture or a continuity. The second one is based on page classification [Gallo et al, 2016], according to the assumption that for two successive pages belong to the same document, they must belong to the same class. After having all the classes of stream pages, consecutive pages that belong to the same class are grouped as a single document.…”

Section: Continuitymentioning

confidence: 99%

“…More recently, one can find in the state of the art deep learning techniques with convolutional neuronal models for the classification of documents, as [Gallo et al, 2016, Harley et al, 2015, Wiedemann and Heyer, 2017. While the first two use only textual information, the last two use textual and visual information.…”

Section: Related Workmentioning

confidence: 99%

Use of Language Models for Document Stream Segmentation

Neche

Belaïd

2020

Proceedings of the 9th International Conference on Pattern Recognition Applications and Methods

View full text Add to dashboard Cite

Page stream segmentation into single documents is a very common task which is practiced in companies and administrations when processing their incoming mail. It is not a straightforward task because the limits of the documents are not always obvious, and it is not always easy to find common features between the pages of the same document. In this paper, we seek to compare existing segmentation models and propose a new segmentation one based on GRUs (Gated Recurrent Unit) and an attention mechanism, named AGRU. This model uses the text content of the previous page and the current page to determine if both pages belong to the same document. So, due to its attention mechanism, this model is capable to recognize words that define the first page of a document. Training and evaluation are carried out on two datasets: Tobacco-800 and READ-Corpus. The former is a public dataset on which our model reaches an F1 score equal to 90%, and the later is private for which our model reaches an F1 score equal to 96%.

show abstract

“…Mais recentemente, uma série de trabalhos [Jain and Wigington 2019, Gallo et al 2016, Wiedemann and Heyer 2019, Audebert et al 2019] propuseram a criação de modelos de segmentação que levam em consideração as informações que estão presentes tanto nas imagens quanto nos textos, em um processo de integração de informações. Os trabalhos supracitados consideraram documentos provenientes de arquivos de grandes empresas, como a base de dados Tobacco800 [Wiedemann and Heyer 2019].…”

unclassified

Classificação de Páginas de Petições iniciais Utilizando Redes Neurais Convolucionais Multimodais

Mota¹,

Lima²,

Nascimento³

et al. 2020

Anais Do Encontro Nacional De Inteligência Artificial E Computacional (ENIAC 2020)

View full text Add to dashboard Cite

A classificação e separação de documentos é uma etapa de extrema importância na análise de processos judiciais. Algoritmos de aprendizado profundo têm alcançado resultados promissores nesta tarefa, extraindo informações relevantes a partir dos textos destes documentos. No entanto, os documentos de processos judiciais têm se tornado cada vez mais heterogêneos, i.e. fotos, recibos, documentos de texto, etc., impactando diretamente a precisão na classificação. Este trabalho investiga o uso de redes convolucionais multimodais, combinando características extraídas de textos e imagens, para classificação de páginas de processos. Duas abordagens multimodais foram comparadas com quatro monomodais. Todos os algoritmos foram avaliados, em termos em acurácia e kappa, em uma base de dados composta por 117 processos judiciais. Os resultados mostraram que a abordagem que atingiu o melhor desempenho é multimodal, apresentando eficácia e eficiência na classificação de páginas de processos.

show abstract

Deep Neural Networks for Page Stream Segmentation and Classification

Cited by 9 publications

References 13 publications

A Multi-Modal Approach to Digital Document Stream Segmentation for Title Insurance Domain

A Multi-Modal Approach to Digital Document Stream Segmentation for Title Insurance Domain

Use of Language Models for Document Stream Segmentation

Classificação de Páginas de Petições iniciais Utilizando Redes Neurais Convolucionais Multimodais

Contact Info

Product

Resources

About