Abstract: This paper provides an overview of the DocILE 2023 Competition, its tasks, participant submissions, the competition results and possible future research directions. This first edition of the competition focused on two Information Extraction tasks, Key Information Localization and Extraction (KILE) and Line Item Recognition (LIR). Both of these tasks require detection of pre-defined categories of information in business documents. The second task additionally requires correctly grouping the information into tuples, capturing the structure laid out in the document. The competition used the recently published DocILE dataset and benchmark that stays open to new submissions. The diversity of the participant solutions indicates the potential of the dataset as the submissions included pure Computer Vision, pure Natural Language Processing, as well as multi-modal solutions and utilized all of the parts of the dataset, including the annotated, synthetic and unlabeled subsets.

Abstract: Presently, man-machine interface development is a widespread research activity. A system to understand hand drawn architectural drawings in a CAD environment is presented in this paper. To understand a document, we have to identify its building elements and their structural properties. An attributed graph structure is chosen as a symbolic representation of the input document and the patterns to recognize in it. An inexact subgraph isomorphism procedure using relaxation labeling techniques is performed. In this paper we focus on how to speed up the matching. There is a building element, the walls, characterized by a hatching pattern. Using a straight line Hough transform (SLHT)-based method, we recognize this pattern, characterized by parallel straight lines, and remove from the input graph the edges belonging to this pattern. The isomorphism is then applied to the remainder of the input graph. When all the building elements have been recognized, the document is redrawn, correcting the inaccurate strokes obtained from a hand-drawn input.

Abstract: In this paper we evaluate the influence of the selection of key points and the associated features in the performance of word spotting processes. In general, features can be extracted from a number of characteristic points like corners, contours, skeletons, maxima, minima, crossings, etc. A number of descriptors exist in the literature using different interest point detectors. But the intrinsic variability of handwriting vary strongly on the performance if the interest points are not stable enough. In this paper, we analyze the performance of different descriptors for local interest points. As benchmarking dataset we have used the Barcelona Marriage Database that contains handwritten records of marriages over five centuries.

Abstract: The importance of logos and trademarks in nowadays society is indisputable, variably seen under a positive light as a valuable service for consumers or a negative one as a catalyst of ever-increasing consumerism. This chapter discusses the technical approaches for enabling machines to work with logos, looking into the latest methodologies for logo detection, localization, representation, recognition, retrieval, and spotting in a variety of media. This analysis is presented in the context of three different applications covering the complete depth and breadth of state of the art techniques. These are trademark retrieval systems, logo recognition in document images, and logo detection and removal in images and videos. This chapter, due to the very nature of logos and trademarks, brings together various facets of document image analysis spanning graphical and textual content, while it links document image analysis to other computer vision domains, especially when it comes to the analysis of real-scene videos and images.

Abstract: The task of image-text matching aims to map representations from different modalities into a common joint visual-textual embedding. However, the most widely used datasets for this task, MSCOCO and Flickr30K, are actually image captioning datasets that offer a very limited set of relationships between images and sentences in their ground-truth annotations. This limited ground truth information forces us to use evaluation metrics based on binary relevance: given a sentence query we consider only one image as relevant. However, many other relevant images or captions may be present in the dataset. In this work, we propose two metrics that evaluate the degree of semantic relevance of retrieved items, independently of their annotated binary relevance. Additionally, we incorporate a novel strategy that uses an image captioning metric, CIDEr, to define a Semantic Adaptive Margin (SAM) to be optimized in a standard triplet loss. By incorporating our formulation to existing models, a large improvement is obtained in scenarios where available training data is limited. We also demonstrate that the performance on the annotated image-caption pairs is maintained while improving on other non-annotated relevant items when employing the full training set. The code for our new metric can be found at github. com/furkanbiten/ncsmetric and the model implementation at github. com/andrespmd/semanticadaptive_margin.