Abstract
We address the challenging problem of Natural Language Comprehension beyond plain-text documents by introducing the TILT neural network architecture which simultaneously learns layout information, visual features, and textual semantics. Contrary to previous approaches, we rely on a decoder capable of unifying a variety of problems involving natural language. The layout is represented as an attention bias and complemented with contextualized visual information, while the core of our model is a pretrained encoder-decoder Transformer. Our novel approach achieves state-of-the-art results in extracting information from documents and answering questions which demand layout understanding (DocVQA, CORD, SROIE). At the same time, we simplify the process by employing an end-to-end model.
R. Powalski, L. Borchmann, D. Jurkiewicz, T. Dwojak and M. Pietruszk—Contributed equally.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
Expected values have always an exact match in CoNLL, but not elsewhere, e.g., it is the case for 20% WikiReading, 27% Kleister, and 93% of SROIE values.
- 2.
- 3.
Per-category test set scores are available after submission on the competition web page: https://rrc.cvc.uab.es/?ch=17&com=evaluation&task=1.
- 4.
Corrections can be obtained by comparing their two public submissions.
References
Cho, M., Amplayo, R., Hwang, S.W., Park, J.: Adversarial TableQA: attention supervision for question answering on tables. In: PMLR (2018)
Choi, E., et al.: QuAC: question answering in context. In: EMNLP (2018)
Chuang, Y., Liu, C., Lee, H., Lee, L.: SpeechBERT: an audio-and-text jointly learned language model for end-to-end spoken question answering. In: ISCA (2020)
Clark, J.H., et al.: TyDi QA: a benchmark for information-seeking question answering in typologically diverse languages. TACL 8, 454–470 (2020)
Dai, J., Li, Y., He, K., Sun, J.: R-FCN: object detection via region-based fully convolutional networks. In: NeurIPS (2016)
Denk, T.I., Reisswig, C.: BERTgrid: contextualized embedding for 2d document representation and understanding. arXiv preprint (2019)
Dodge, J., Ilharco, G., Schwartz, R., Farhadi, A., Hajishirzi, H., Smith, N.A.: Fine-tuning pretrained language models: weight initializations, data orders, and early stopping. arXiv preprint (2020)
Dua, D., Wang, Y., Dasigi, P., Stanovsky, G., Singh, S., Gardner, M.: DROP: a reading comprehension benchmark requiring discrete reasoning over paragraphs. In: NAACL-HLT (2019)
Dwojak, T., Pietruszka, M., Borchmann, Ł., Chłedowski, J., Graliński, F.: From dataset recycling to multi-property extraction and beyond. In: CoNLL (2020)
Ethayarajh, K.: How contextual are contextualized word representations? comparing the geometry of BERT, ELMo, and GPT-2 embeddings. In: EMNLP-IJCNLP (2019)
Garncarek, Ł., et al.: LAMBERT: layout-aware (language) modeling using BERT for information extraction. In: Llads, J. et al. (eds.) ICDAR 2021. LNCS, vol. 12822, pp. 532–547 (2021). Accepted to ICDAR 2021
Guu, K., Lee, K., Tung, Z., Pasupat, P., Chang, M.: Retrieval augmented language model pre-training. In: ICML (2020)
Han, K., et al.: A survey on visual transformer. arXiv preprint (2021)
Harley, A.W., Ufkes, A., Derpanis, K.G.: Evaluation of deep convolutional nets for document image classification and retrieval. In: ICDAR (2015)
Herzig, J., Nowak, P.K., Müller, T., Piccinno, F., Eisenschlos, J.: TaPas: weakly supervised table parsing via pre-training. In: ACL (2020)
Hewlett, D., et al.: WikiReading: a novel large-scale language understanding task over Wikipedia. In: ACL (2016)
Ho, J., Kalchbrenner, N., Weissenborn, D., Salimans, T.: Axial attention in multidimensional transformers. arXiv preprint (2019)
Hong, T., Kim, D., Ji, M., Hwang, W., Nam, D., Park, S.: BROS: a pre-trained language model for understanding texts in document. openreview.net preprint (2021)
Huang, Z., et al.: ICDAR2019 competition on scanned receipt OCR and information extraction. In: ICDAR (2019)
Hwang, W., Yim, J., Park, S., Yang, S., Seo, M.: Spatial dependency parsing for semi-structured document information extraction. arXiv preprint (2020)
Jaume, G., Ekenel, H.K., Thiran, J.P.: FUNSD: a dataset for form understanding in noisy scanned documents. In: ICDAR-OST (2019)
Kafle, K., Price, B.L., Cohen, S., Kanan, C.: DVQA: understanding data visualizations via question answering. In: CVPR (2018)
Kahou, S.E., Michalski, V., Atkinson, A., Kádár, Á., Trischler, A., Bengio, Y.: FigureQA: an annotated figure dataset for visual reasoning. In: ICLR (2018)
Kasai, J., Pappas, N., Peng, H., Cross, J., Smith, N.A.: Deep encoder, shallow decoder: reevaluating the speed-quality tradeoff in machine translation. arXiv preprint (2020)
Keskar, N., McCann, B., Xiong, C., Socher, R.: Unifying question answering and text classification via span extraction. arXiv preprint (2019)
Khashabi, D., et al.: UnifiedQA: crossing format boundaries with a single QA system. In: EMNLP-Findings (2020)
Khot, T., Clark, P., Guerquin, M., Jansen, P., Sabharwal, A.: QASC: a dataset for question answering via sentence composition. In: AAAI (2020)
Kudo, T.: Subword regularization: improving neural network translation models with multiple subword candidates. In: ACL (2018)
Kumar, A., et al.: Ask me anything: dynamic memory networks for natural language processing. In: ICML (2016)
Kwiatkowski, T., et al.: Natural questions: a benchmark for question answering research. TACL 7, 453–466 (2019)
Lai, G., Xie, Q., Liu, H., Yang, Y., Hovy, E.: RACE: large-scale ReAding comprehension dataset from examinations. In: EMNLP (2017)
Le, H., Sahoo, D., Chen, N., Hoi, S.: Multimodal transformer networks for end-to-end video-grounded dialogue systems. In: ACL (2019)
Lee, K.-H., Chen, X., Hua, G., Hu, H., He, X.: Stacked cross attention for image-text matching. In: Ferrari, V., Hebert, M., Sminchisescu, C., Weiss, Y. (eds.) ECCV 2018. LNCS, vol. 11208, pp. 212–228. Springer, Cham (2018). https://doi.org/10.1007/978-3-030-01225-0_13
Lewis, M., et al.: BART: denoising sequence-to-sequence pre-training for natural language generation, translation, and comprehension. In: ACL (2020)
Li, L.H., Yatskar, M., Yin, D., Hsieh, C.J., Chang, K.W.: VisualBERT: a simple and performant baseline for vision and language. arXiv preprint (2019)
Liu, X., Gao, F., Zhang, Q., Zhao, H.: Graph convolution for multimodal information extraction from visually rich documents. In: NAACL-HLT (2019)
Ma, J., Qin, S., Su, L., Li, X., Xiao, L.: Fusion of image-text attention for transformer-based multimodal machine translation. In: IALP (2019)
Mathew, M., Karatzas, D., Jawahar, C.: DocVQA: a dataset for VQA on document images. In: WACV, pp. 2200–2209 (2021)
McCann, B., Keskar, N.S., Xiong, C., Socher, R.: The natural language decathlon: multitask learning as question answering. arXiv preprint (2018)
Palm, R.B., Winther, O., Laws, F.: CloudScan - a configuration-free invoice analysis system using recurrent neural networks. In: ICDAR (2017)
Park, S., et al.: CORD: a consolidated receipt dataset for post-OCR parsing. In: Document Intelligence Workshop at NeurIPS (2019)
Powalski, R., Stanislawek, T.: UniCase - rethinking casing in language models. arXiv prepint (2020)
Radford, A., Wu, J., Child, R., Luan, D., Amodei, D., Sutskever, I.: Language models are unsupervised multitask learners. Technical report (2019)
Raffel, C., et al.: Exploring the limits of transfer learning with a unified text-to-text transformer. JMRL (2020)
Rajpurkar, P., Zhang, J., Lopyrev, K., Liang, P.: SQuAD: 100,000+ questions for machine comprehension of text. In: EMNLP (2016)
Reddy, S., Chen, D., Manning, C.D.: CoQA: a conversational question answering challenge. TACL 7, 249–266 (2019)
Ren, Y., Liu, J., Tan, X., Zhao, Z., Zhao, S., Liu, T.Y.: A study of non-autoregressive model for sequence generation. In: ACL (2020)
Ronneberger, O., Fischer, P., Brox, T.: U-Net: convolutional networks for biomedical image segmentation. In: Navab, N., Hornegger, J., Wells, W.M., Frangi, A.F. (eds.) MICCAI 2015. LNCS, vol. 9351, pp. 234–241. Springer, Cham (2015). https://doi.org/10.1007/978-3-319-24574-4_28
Sennrich, R., Haddow, B., Birch, A.: Neural machine translation of rare words with subword units. In: ACL (2016)
Sidorov, O., Hu, R., Rohrbach, M., Singh, A.: TextCaps: a dataset for image captioning with reading comprehension. In: Vedaldi, A., Bischof, H., Brox, T., Frahm, J.-M. (eds.) ECCV 2020. LNCS, vol. 12347, pp. 742–758. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-58536-5_44
Singh, A., et al.: Towards VQA models that can read. In: CVPR (2019)
Stanisławek, T., et al.: Kleister: key information extraction datasets involving long documents with complex layouts. In: Llads, J. et al. (eds.) ICDAR 2021. LNCS, vol. 12822, pp. 564–579 (2021). Accepted to ICDAR 2021
Su, W., Zhu, X., Cao, Y., Li, B., Lu, L., Wei, F., Dai, J.: VL-BERT: pre-training of generic visual-linguistic representations. In: ICLR (2020)
Vaswani, A., et al.: Attention is all you need. In: NeurIPS (2017)
Xu, Y., et al.: LayoutLMv2: multi-modal pre-training for visually-rich document understanding. arXiv preprint (2020)
Xu, Y., Li, M., Cui, L., Huang, S., Wei, F., Zhou, M.: LayoutLM: pre-training of text and layout for document image understanding. In: KDD (2020)
Yin, P., Neubig, G., Yih, W.t., Riedel, S.: TaBERT: pretraining for joint understanding of textual and tabular data. In: ACL (2020)
Acknowledgments
The authors would like to thank Filip Graliński, Tomasz Stanisławek, and Łukasz Garncarek for fruitful discussions regarding the paper and our managing directors at Applica.ai. Moreover, Dawid Jurkiewicz pays due thanks to his son for minding the deadline and generously coming into the world a day after.
The Smart Growth Operational Programme supported this research under project no. POIR.01.01.01-00-0877/19-00 (A universal platform for robotic automation of processes requiring text comprehension, with a unique level of implementation and service automation).
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this paper
Cite this paper
Powalski, R., Borchmann, Ł., Jurkiewicz, D., Dwojak, T., Pietruszka, M., Pałka, G. (2021). Going Full-TILT Boogie on Document Understanding with Text-Image-Layout Transformer. In: Lladós, J., Lopresti, D., Uchida, S. (eds) Document Analysis and Recognition – ICDAR 2021. ICDAR 2021. Lecture Notes in Computer Science(), vol 12822. Springer, Cham. https://doi.org/10.1007/978-3-030-86331-9_47
Download citation
DOI: https://doi.org/10.1007/978-3-030-86331-9_47
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-86330-2
Online ISBN: 978-3-030-86331-9
eBook Packages: Computer ScienceComputer Science (R0)