Abstract: Graph representations have been widely used in pattern recognition thanks to their powerful representation formalism and rich theoretical background. A number of error-tolerant graph matching algorithms such as graph edit distance have been proposed for computing a distance between two labelled graphs. However, they typically suffer from a high
computational complexity, which makes it difficult to apply
these matching algorithms in a real scenario. In this paper, we propose an efficient graph distance based on the emerging field of geometric deep learning. Our method employs a message passing neural network to capture the graph structure and learns a metric with a siamese network approach. The performance of the proposed graph distance is validated in two application cases, graph classification and graph retrieval of handwritten words, and shows a promising performance when compared with
(approximate) graph edit distance benchmarks.

Abstract: When extracting information from handwritten documents, text transcription and named entity recognition are usually faced as separate subsequent tasks. This has the disadvantage that errors in the first module affect heavily the
performance of the second module. In this work we propose to do both tasks jointly, using a single neural network with a common architecture used for plain text recognition. Experimentally, the work has been tested on a collection of historical marriage records. Results of experiments are presented to show the effect on the performance for different
configurations: different ways of encoding the information, doing or not transfer learning and processing at text line or multi-line region level. The results are comparable to state of the art reported in the ICDAR 2017 Information Extraction competition, even though the proposed technique does not use any dictionaries, language modeling or post processing.

Abstract: The immense success of deep learning based methods in computer vision heavily relies on large scale training datasets. These richly annotated datasets help the network learn discriminative visual features. Collecting and annotating such datasets requires a tremendous amount of human effort and annotations are limited to popular set of classes. As an alternative, learning visual features by designing auxiliary tasks which make use of freely available self-supervision has become increasingly popular in the computer vision community.
In this paper, we put forward an idea to take advantage of multi-modal context to provide self-supervision for the training of computer vision algorithms. We show that adequate visual features can be learned efficiently by training a CNN to predict the semantic textual context in which a particular image is more probable to appear as an illustration. More specifically we use popular text embedding techniques to provide the self-supervision for the training of deep CNN.

Abstract: With the ongoing progress in digitization, huge document collections and archives have become available to a broad audience. Scanned document images can be transmitted electronically and studied simultaneously throughout the world. While this is very beneficial, it is often impossible to perform automated searches on these document collections. Optical character recognition usually fails when it comes to handwritten or historic documents. In order to address the need for exploring document collections rapidly, researchers are working on word spotting. In query-by-example word spotting scenarios, the user selects an exemplary occurrence of the query word in a document image. The word spotting system then retrieves all regions in the collection that are visually similar to the given example of the query word. The best matching regions are presented to the user and no actual transcription is required.
An important property of a word spotting system is the computational speed with which queries can be executed. In our previous work, we presented a relatively slow but high-precision method. In the present work, we will extend this baseline system to an integrated two-stage approach. In a coarse-grained first stage, we will filter document images efficiently in order to identify regions that are likely to contain the query word. In the fine-grained second stage, these regions will be analyzed with our previously presented high-precision method. Finally, we will report recognition results and query times for the well-known George Washington
benchmark in our evaluation. We achieve state-of-the-art recognition results while the query times can be reduced to 50% in comparison with our baseline.

Abstract: Tabular structures in business documents offer a complementary dimension to the raw textual data. For instance, there is information about the relationships among pieces of information. Nowadays, digital mailroom applications have become a key service for workflow automation. Therefore, the detection and interpretation of tables is crucial. With the recent advances in information extraction, table detection and recognition has gained interest in document image analysis, in particular, with the absence of rule lines and unknown information about rows and columns. However, business documents usually contain sensitive contents limiting the amount of public benchmarking datasets. In this paper, we propose a graph-based approach for detecting tables in document images which do not require the raw content of the document. Hence, the sensitive content can be previously removed and, instead of using the raw image or textual content, we propose a purely structural approach to keep sensitive data anonymous. Our framework uses graph neural networks (GNNs) to describe the local repetitive structures that constitute a table. In particular, our main application domain are business documents. We have carefully validated our approach in two invoice datasets and a modern document benchmark. Our experiments demonstrate that tables can be detected by purely structural approaches.