Abstract: Sectors as fintech, legaltech or insurance process an inflow of millions of forms, invoices, id documents, claims or similar every day. Together with these, historical archives provide gigantic amounts of digitized documents containing useful information that needs to be stored in machine encoded text with a meaningful structure. This procedure, known as information extraction (IE) comprises the steps of localizing and recognizing text, identifying named entities contained in it and optionally finding relationships among its elements. In this work we explore multi-task neural models at image and graph level to solve all steps in a unified way. While doing so we find benefits and limitations of these end-to-end approaches in comparison with sequential separate methods. More specifically, we first propose a method to produce textual as well as semantic labels with a unified model from handwritten text line images. We do so with the use of a convolutional recurrent neural model trained with connectionist temporal classification to predict the textual as well as semantic information encoded in the images. Secondly, motivated by the success of this approach we investigate the unification of the localization and recognition tasks of handwritten text in full pages with an end-to-end model, observing benefits in doing so. Having two models that tackle information extraction subsequent task pairs in an end-to-end to end manner, we lastly contribute with a method to put them all together in a single neural network to solve the whole information extraction pipeline in a unified way. Doing so we observe some benefits and some limitations in the approach, suggesting that in certain cases it is beneficial to train specialized models that excel at a single challenging task of the information extraction process, as it can be the recognition of named entities or the extraction of relationships between them. For this reason we lastly study the use of the recently arrived graph neural network architectures for the semantic tasks of the information extraction process, which are recognition of named entities and relation extraction, achieving promising results on the relation extraction part.

Abstract: In this paper we present a method to optimize the topology of linear left-to-right hidden Markov models. These models are very popular for sequential signals modeling on tasks such as handwriting recognition. Many topology definition methods select the number of states for a character model based
on character length. This can be a drawback when characters are shorter than the minimum allowed by the model, since they can not be properly trained nor recognized. The proposed method optimizes the number of states per model by automatically including convenient skip-state transitions and therefore it avoids the aforementioned problem.We discuss and compare our method with other character length-based methods such the Fixed, Bakis and Quantile methods. Our proposal performs well on off-line handwriting recognition task.

Abstract: As mobile document acquisition using smartphones is getting more and more common, along with the continuous improvement of mobile devices (both in terms of computing power and image quality), we can wonder to which extent mobile phones can replace desktop scanners. Modern applications can cope with perspective distortion and normalize the contrast of a document page captured with a smartphone, and in some cases like bottle labels or posters, smartphones even have the advantage of allowing the acquisition of non-flat or large documents. However, several cases remain hard to handle, such as reflective documents (identity cards, badges, glossy magazine cover, etc.) or large documents for which some regions require an important amount of detail. This paper introduces the SmartDoc 2017 benchmark (named “SmartDoc Video Capture”), which aims at
assessing whether capturing documents using the video mode of a smartphone could solve those issues. The task under evaluation is both a stitching and a reconstruction problem, as the user can move the device over different parts of the document to capture details or try to erase highlights. The material released consists of a dataset, an evaluation method and the associated tool, a sample method, and the tools required to extend the dataset. All the components are released publicly under very permissive licenses, and we particularly cared about maximizing the ease of
understanding, usage and improvement.

Abstract: This note introduces a visual attention model of text localization in real-world scenes. The core of the model built upon the proto-object concept is discussed. It is shown how such dynamic mid-level representation of the scene can be derived in the framework of an action-perception loop engaging salience, text information value computation, and eye guidance mechanisms.
Preliminary results that compare model generated scanpaths with those eye-tracked from human subjects are presented.

Abstract: Searching for text objects in real scene images is an open problem and a very active computer vision research area. A large number of methods have been proposed tackling the text search as extension of the ones from the document analysis field or inspired by general purpose object detection methods. However the general problem of object search in real scene images remains an extremely challenging problem due to the huge variability in object appearance. This thesis builds on top of the most recent findings in the visual attention literature presenting a novel computational model of eye guidance aiming to better describe text object search in real scene images.
First are presented the relevant state-of-the-art results from the visual attention literature regarding eye movements and visual search. Relevant models of attention are discussed and integrated with recent observations on the role of top-down constraints and the emerging need for a layered model of attention in which saliency is not the only factor guiding attention. Visual attention is then explained by the interaction of several modulating factors, such as objects, value, plans and saliency. Then we introduce our probabilistic formulation of attention deployment in real scene. The model is based on the rationale that oculomotor control depends on two interacting but distinct processes: an attentional process that assigns value to the sources of information and motor process that flexibly links information with action.
In such framework, the choice of where to look next is task-dependent and oriented to classes of objects embedded within pictures of complex scenes. The dependence on task is taken into account by exploiting the value and the reward of gazing at certain image patches or proto-objects that provide a sparse representation of the scene objects.
In the experimental section the model is tested in laboratory condition, comparing model simulations with data from eye tracking experiments. The comparison is qualitative in terms of observable scan paths and quantitative in terms of statistical similarity of gaze shift amplitude. Experiments are performed using eye tracking data from both a publicly available dataset of face and text and from newly performed eye-tracking experiments on a dataset of street view pictures containing text. The last part of this thesis is dedicated to study the extent to which the proposed model can account for human eye movements in a low constrained setting. We used a mobile eye tracking device and an ad-hoc developed methodology to compare model simulated eye data with the human eye data from mobile eye tracking recordings. Such setting allow to test the model in an incomplete visual information condition, reproducing a close to real-life search task.