Rostand Costa

A Study on the Use of Multiple Avatars in 3D Sign Language Dictionaries

Numerous platforms in the field of machine translation of oralized languages to sign language are available nowadays, and accessibility has been gaining more and more space. However, it is noticed that most platforms use only a unique 3D avatar, and this character is responsible for all the reproduction of signals, with no alternative of choice for users. Such a limitation may have an impact on the acceptance of automatic translation by the deaf community, since there must be empathy of the deaf with the animated agent. Having only one available avatar makes impossible a more precise choice, which may involve personal characteristics and affinities. One of the reasons for this is the great effort, human and technological, that is necessary for the construction of a sign dictionary, which can scale proportionally with the addition of new avatars. In view of such a scenario, the present study aims to investigate mechanisms that allow multiple avatars to be offered in sign dictionaries without necessarily needing to reshape them again and manually, one by one. The initial premise is to analyze the functioning of each signal in a particular avatar, in order to predict possible problems in the reproduction of the signals after the permutation to a new one (retargeting), such as improper collisions or mesh invasions. As main contributions of the work, techniques are proposed to facilitate the identification and automatic correction of nonconformities in the movement of the signals and also some practical recommendations for the modeling of new avatars in order to minimize the occurrence of errors.

Urban data collectors: A pragmatic approach to leveraging urban sensing

This paper proposes the adoption of a more pragmatic approach to urban sensing to leverage the initial offering of open databases within the context of smart cities. In this sense, the use of mobility agents already available to provide mobility to traditional sensors can enable a preliminary instrumentation of cities more quickly and possibly at a lower cost. As proof of concept, we investigated the use of metropolitan bus equipped with sensors such as urban data collectors. We believe that this strategy can help overcoming many challenges of traditional approaches for instrumentation of cities, such as power supply, maintenance, operation, and protection of a large amount of static sensors. To assess the feasibility of sampling by collectors, we conducted experiments comparing averages using frequent readings vs averages using intermittent readings collected by sensors of a climatological micro station.

VLibras-BOX: Flexible Portuguese-LIBRAS Translation Server Encapsulation for Distributed, Centralized or Hybrid Scenarios

In this paper we present the concept of a VLibras-Box, a building box for distributed sign-language translation infrastructures. A VLibras-Box encapsulates a Portuguese to Brazilian Sign Language translation service and additional components needed to combine multiple VLibras-Boxes in order to assembly a fault-tolerant and scalable translation infrastructure. We present three different application scenarios, hanging from a single VLibras-Box, where the translation endpoint is defined in an static way to multiple translation servers scenarios with dynamic selection and a two-way load balancing strategy to compose public, private and hybrid fault-tolerant distributed translation infrastructures.

A Study On The Use of Deep Learning for Automatic Audience Counting

Counting objects or living beings is a common necessity in many areas of industry, commerce and services. Automating this activity can promote an optimization of the process involved and, consequently, the reduction of time and costs. With this in mind, computer vision is an approach that provides new possibilities for the digital processing of images, giving the computer a capacity of interpretation increasingly similar to humans. This work aims to compare the efficiency of volumetric counting techniques, both using traditional computational vision and deep learning, in counting audiences in face-to-face events. As a case study, this preliminary investigation focused on audience counting of film and / or theater sessions from audience photos. Gauge billing automatically, accurately and transparently is a recurring need of the entertainment industry. From our experiments it was possible to observe the great potential of the application of deep learning in this context. When compared to several automatic volumetric counting techniques available, deep learning was the strategy that presented the best results, reaching sensitivity and precision above 96%.

Asynchronous Queue Based Approach for Building Reactive Microservices

To achieve scalability and flexibility in larger applications a new approach arises, named by Microservices (MS). However MS architectures are at their inception and are even more a concept than a fully mature design pattern. One of the hardest topics in this approach is how to properly migrate or develop a single microservice, in terms of scope, efficiency and dependability. In this sense, this work proposes a new architectural model based on high-level architecture pattern of reactive programming to the internal structure of a new microservice. The new model of microservices are internally coordinated by asynchronous queues, which allowed to preserve compatibility with most monolithic components and provide an encapsulation process to enable its continuity. A comparative study between the standard approach and the proposed architecture was carried out to measure the eventual performance improvement of the new strategy.

A Flexible Architecture for Building Ruled-Based Text-to-Gloss Machine Translators: Extended Abstract

Deaf people face a number of difficulties in accessing information, primarily in dynamic, high-volume environments such as the Web. One of the major obstacles is due to the fact that this information is available almost entirely in the oral language only, while the deaf communicate naturally using signals. There are some solutions available in the literature for automatic translation of oral language content into sign languages, but they are usually restricted to translating content into a single input oral language. The objective of this work is to investigate the feasibility of generalization of a translation strategy with syntactic-semantic treatment, originally directioned from Brazilian Portuguese for LIBRAS, to define a flexible architecture for the construction of automatic translators of several oral languages for associated sign languages. The hypothesis we wish to evaluate is that an extensible and configurable translation core, based on the interpretation of formally defined rules and grammars acting at the morphological, syntactic or semantic levels, can be used to implement new machine translation initiatives for sign languages or as a way to improve the quality of translation in existing tools.

An Open and Extensible Platform for Machine Translation of Spoken Languages into Sign Languages

The purpose of this paper is to investigate the feasibility of offering a multilingual platform for text-to-sign translation, i.e., a solution where a machine translates digital contents in several spoken languages to several sign languages in scenarios such as Digital TV, Web and Cinema. This solution called OpenSigns, is an open platform that has several common components for generic functionalities, initially originated from the Suite VLibras, including the creation and manipulation of 3D animation models, and interchangeable mechanisms specific for each sign language, such as a text-to-gloss machine translation engine, a sign dictionary for each sign language, among others. Our motivation is that the concentration of efforts and resources around a single solution could provide some state-of-the-art improvement, such as a standard solution for the industry and a greater functional flexibility for common components. In addition, we could also share techniques and heuristics between the translation mechanisms, reducing the effort to make a new sign language available on the platform, which may further enhance digital inclusion and accessibility, especially for poorest countries.

We Need to Talk About Digital Preservation of Audiovisual Collections: Strategies for Building National Networks

This paper is about the relevance of creating national networks to digital preservation of audiovisual collections, from both film and television industry. The proposal arises from the perception that, in the same proportion that the audiovisual production is being created in digital formats, it is expected the preservation of such information to be based on appropriate techniques and technologies also digital. It is similarly considered that the concern of digitization and preservation of collections are present in several initiatives guiding public policy in many countries. Based on the experiments already undergoing in the world, especially in the most advanced countries in the area, it is possible to predict that the building of a national network for the promotion of audiovisual digital preservation, with range and broad scope, can significantly contribute to the development of the national expertise in the area, to consolidate partnerships and to leverage the emergence of solutions adapted to the context of each country. Steps in this direction could include, among other actions, the promotion of two complementary strands of initiatives: (a) the organization of a consortium of public and private organizations interested in digital preservation; and (b) assembling a “backbone” of national repositories for large-scale digital preservation.

QD Framework: Using Dynamic Catalogs to Organize Mirror Servers on Scalable and Resilient Service Meshes

In this paper we conceptualize and define the QD Framework, a generic solution to manage a set of mirror servers, enabling a specific online service to be offered on the web with a better availability and success rate for requests. The framework includes all the components needed to deal with such service: registration, cataloging, monitoring, distribution and access for the end users. For validation, a case study was done with an implementation of the proposed framework to support a large scale service that provides machine translation of digital content in portuguese language to LIBRAS (Brazilian Sign Language).

Using video embedded markings for supporting content sensitive interaction in multiple contexts

Despite the numerous advances provided by Digital TV and Connected TV platforms, they have not provided significant gains to the television viewing experience. In the Connected TV platform, for instance, one relevant limitation is that the audiovisual content normally does not converse with the resources of these environments. In this work we propose a complementary approach to promote content-sensitive interactivity. This approach uses markup on video to promote content-sensitive interactivity based on self-contained hypermedia anchors. The anchors are self-contained metadata inserted synchronously with the narrative and contain necessary information to allow that a specific agent can lead the users to different contexts according to their interaction with the content. Our experiments show that is possible to create rich interactive narratives even with a limited set of anchors classes, including support for pull, push and distributed modes.