904 resultados para Audio-Visual Automatic Speech Recognition
Resumo:
[ES] Los erizos de mar han servido como modelo prototípico de organismo en el desarrollo de la Biología. La irrupción de este animal como especie invasora en los fondos canarios, combinada con el éxito reproductivo que ha tenido en nuestras aguas, ha creado un problema medioambiental importante que se ha intentado atajar con la puesta en marcha de proyectos e iniciativas orientados a su erradicación (matanzas masivas) o su contención con intentos de estimular su explotación comercial para uso gastronómico. En el transcurso de este trabajo se pretende explorar la robustez con la que se pueden clasificar visualmente diferentes tipos de erizos (principalmente Diadema antillarumy y Erizos autóctonos) a partir tanto de imágenes estáticas como de secuencias de vídeo para evaluar si, mediante el empleo de técnicas de visión por computador, es posible resolver estas tareas mediante la inspección automática de vídeos e imágenes.
Resumo:
[EN]An accurate estimation of the number of people entering / leaving a controlled area is an interesting capability for automatic surveil- lance systems. Potential applications where this technology can be ap- plied include those related to security, safety, energy saving or fraud control. In this paper we present a novel con guration of a multi-sensor system combining both visual and range data specially suited for trou- blesome scenarios such as public transportation. The approach applies probabilistic estimation lters on raw sensor data to create intermediate level hypothesis that are later fused using a certainty-based integration stage. Promising results have been obtained in several tests performed on a realistic test bed scenario under variable lightning conditions.
Resumo:
The automatic extraction of biometric descriptors of anonymous people is a challenging scenario in camera networks. This task is typically accomplished making use of visual information. Calibrated RGBD sensors make possible the extraction of point cloud information. We present a novel approach for people semantic description and re-identification using the individual point cloud information. The proposal combines the use of simple geometric features with point cloud features based on surface normals.
Resumo:
[EN]During the last decade, researchers have verified that clothing can provide information for gender recognition. However, before extracting features, it is necessary to segment the clothing region. We introduce a new clothes segmentation method based on the application of the GrabCut technique over a trixel mesh, obtaining very promising results for a close to real time system. Finally, the clothing features are combined with facial and head context information to outperform previous results in gender recognition with a public database.
Resumo:
Process algebraic architectural description languages provide a formal means for modeling software systems and assessing their properties. In order to bridge the gap between system modeling and system im- plementation, in this thesis an approach is proposed for automatically generating multithreaded object-oriented code from process algebraic architectural descriptions, in a way that preserves – under certain assumptions – the properties proved at the architectural level. The approach is divided into three phases, which are illustrated by means of a running example based on an audio processing system. First, we develop an architecture-driven technique for thread coordination management, which is completely automated through a suitable package. Second, we address the translation of the algebraically-specified behavior of the individual software units into thread templates, which will have to be filled in by the software developer according to certain guidelines. Third, we discuss performance issues related to the suitability of synthesizing monitors rather than threads from software unit descriptions that satisfy specific constraints. In addition to the running example, we present two case studies about a video animation repainting system and the implementation of a leader election algorithm, in order to summarize the whole approach. The outcome of this thesis is the implementation of the proposed approach in a translator called PADL2Java and its integration in the architecture-centric verification tool TwoTowers.
Resumo:
[EN]Different researches suggest that inner facial features are not the only discriminative features for tasks such as person identification or gender classification. Indeed, they have shown an influence of features which are part of the local face context, such as hair, on these tasks. However, object-centered approaches which ignore local context dominate the research in computational vision based facial analysis. In this paper, we performed an analysis to study which areas and which resolutions are diagnostic for the gender classification problem. We first demonstrate the importance of contextual features in human observers for gender classification using a psychophysical ”bubbles” technique.
Resumo:
[EN]In this paper a system for face recognition from a tabula rasa (i.e. blank slate) perspective is described. A priori, the system has the only ability to detect automatically faces and represent them in a space of reduced dimension. Later, the system is exposed to over 400 different identities, observing its recognition performance evolution. The preliminary results achieved indicate on the one side that the system is able to reject most of unknown individuals after an initialization stage.
Resumo:
Visual correspondence is a key computer vision task that aims at identifying projections of the same 3D point into images taken either from different viewpoints or at different time instances. This task has been the subject of intense research activities in the last years in scenarios such as object recognition, motion detection, stereo vision, pattern matching, image registration. The approaches proposed in literature typically aim at improving the state of the art by increasing the reliability, the accuracy or the computational efficiency of visual correspondence algorithms. The research work carried out during the Ph.D. course and presented in this dissertation deals with three specific visual correspondence problems: fast pattern matching, stereo correspondence and robust image matching. The dissertation presents original contributions to the theory of visual correspondence, as well as applications dealing with 3D reconstruction and multi-view video surveillance.
Resumo:
The construction and use of multimedia corpora has been advocated for a while in the literature as one of the expected future application fields of Corpus Linguistics. This research project represents a pioneering experience aimed at applying a data-driven methodology to the study of the field of AVT, similarly to what has been done in the last few decades in the macro-field of Translation Studies. This research was based on the experience of Forlixt 1, the Forlì Corpus of Screen Translation, developed at the University of Bologna’s Department of Interdisciplinary Studies in Translation, Languages and Culture. As a matter of fact, in order to quantify strategies of linguistic transfer of an AV product, we need to take into consideration not only the linguistic aspect of such a product but all the meaning-making resources deployed in the filmic text. Provided that one major benefit of Forlixt 1 is the combination of audiovisual and textual data, this corpus allows the user to access primary data for scientific investigation, and thus no longer rely on pre-processed material such as traditional annotated transcriptions. Based on this rationale, the first chapter of the thesis sets out to illustrate the state of the art of research in the disciplinary fields involved. The primary objective was to underline the main repercussions on multimedia texts resulting from the interaction of a double support, audio and video, and, accordingly, on procedures, means, and methods adopted in their translation. By drawing on previous research in semiotics and film studies, the relevant codes at work in visual and acoustic channels were outlined. Subsequently, we concentrated on the analysis of the verbal component and on the peculiar characteristics of filmic orality as opposed to spontaneous dialogic production. In the second part, an overview of the main AVT modalities was presented (dubbing, voice-over, interlinguistic and intra-linguistic subtitling, audio-description, etc.) in order to define the different technologies, processes and professional qualifications that this umbrella term presently includes. The second chapter focuses diachronically on various theories’ contribution to the application of Corpus Linguistics’ methods and tools to the field of Translation Studies (i.e. Descriptive Translation Studies, Polysystem Theory). In particular, we discussed how the use of corpora can favourably help reduce the gap existing between qualitative and quantitative approaches. Subsequently, we reviewed the tools traditionally employed by Corpus Linguistics in regard to the construction of traditional “written language” corpora, to assess whether and how they can be adapted to meet the needs of multimedia corpora. In particular, we reviewed existing speech and spoken corpora, as well as multimedia corpora specifically designed to investigate Translation. The third chapter reviews Forlixt 1's main developing steps, from a technical (IT design principles, data query functions) and methodological point of view, by laying down extensive scientific foundations for the annotation methods adopted, which presently encompass categories of pragmatic, sociolinguistic, linguacultural and semiotic nature. Finally, we described the main query tools (free search, guided search, advanced search and combined search) and the main intended uses of the database in a pedagogical perspective. The fourth chapter lists specific compilation criteria retained, as well as statistics of the two sub-corpora, by presenting data broken down by language pair (French-Italian and German-Italian) and genre (cinema’s comedies, television’s soapoperas and crime series). Next, we concentrated on the discussion of the results obtained from the analysis of summary tables reporting the frequency of categories applied to the French-Italian sub-corpus. The detailed observation of the distribution of categories identified in the original and dubbed corpus allowed us to empirically confirm some of the theories put forward in the literature and notably concerning the nature of the filmic text, the dubbing process and Italian dubbed language’s features. This was possible by looking into some of the most problematic aspects, like the rendering of socio-linguistic variation. The corpus equally allowed us to consider so far neglected aspects, such as pragmatic, prosodic, kinetic, facial, and semiotic elements, and their combination. At the end of this first exploration, some specific observations concerning possible macrotranslation trends were made for each type of sub-genre considered (cinematic and TV genre). On the grounds of this first quantitative investigation, the fifth chapter intended to further examine data, by applying ad hoc models of analysis. Given the virtually infinite number of combinations of categories adopted, and of the latter with searchable textual units, three possible qualitative and quantitative methods were designed, each of which was to concentrate on a particular translation dimension of the filmic text. The first one was the cultural dimension, which specifically focused on the rendering of selected cultural references and on the investigation of recurrent translation choices and strategies justified on the basis of the occurrence of specific clusters of categories. The second analysis was conducted on the linguistic dimension by exploring the occurrence of phrasal verbs in the Italian dubbed corpus and by ascertaining the influence on the adoption of related translation strategies of possible semiotic traits, such as gestures and facial expressions. Finally, the main aim of the third study was to verify whether, under which circumstances, and through which modality, graphic and iconic elements were translated into Italian from an original corpus of both German and French films. After having reviewed the main translation techniques at work, an exhaustive account of possible causes for their non-translation was equally provided. By way of conclusion, the discussion of results obtained from the distribution of annotation categories on the French-Italian corpus, as well as the application of specific models of analysis allowed us to underline possible advantages and drawbacks related to the adoption of a corpus-based approach to AVT studies. Even though possible updating and improvement were proposed in order to help solve some of the problems identified, it is argued that the added value of Forlixt 1 lies ultimately in having created a valuable instrument, allowing to carry out empirically-sound contrastive studies that may be usefully replicated on different language pairs and several types of multimedia texts. Furthermore, multimedia corpora can also play a crucial role in L2 and translation teaching, two disciplines in which their use still lacks systematic investigation.
Resumo:
Images of a scene, static or dynamic, are generally acquired at different epochs from different viewpoints. They potentially gather information about the whole scene and its relative motion with respect to the acquisition device. Data from different (in the spatial or temporal domain) visual sources can be fused together to provide a unique consistent representation of the whole scene, even recovering the third dimension, permitting a more complete understanding of the scene content. Moreover, the pose of the acquisition device can be achieved by estimating the relative motion parameters linking different views, thus providing localization information for automatic guidance purposes. Image registration is based on the use of pattern recognition techniques to match among corresponding parts of different views of the acquired scene. Depending on hypotheses or prior information about the sensor model, the motion model and/or the scene model, this information can be used to estimate global or local geometrical mapping functions between different images or different parts of them. These mapping functions contain relative motion parameters between the scene and the sensor(s) and can be used to integrate accordingly informations coming from the different sources to build a wider or even augmented representation of the scene. Accordingly, for their scene reconstruction and pose estimation capabilities, nowadays image registration techniques from multiple views are increasingly stirring up the interest of the scientific and industrial community. Depending on the applicative domain, accuracy, robustness, and computational payload of the algorithms represent important issues to be addressed and generally a trade-off among them has to be reached. Moreover, on-line performance is desirable in order to guarantee the direct interaction of the vision device with human actors or control systems. This thesis follows a general research approach to cope with these issues, almost independently from the scene content, under the constraint of rigid motions. This approach has been motivated by the portability to very different domains as a very desirable property to achieve. A general image registration approach suitable for on-line applications has been devised and assessed through two challenging case studies in different applicative domains. The first case study regards scene reconstruction through on-line mosaicing of optical microscopy cell images acquired with non automated equipment, while moving manually the microscope holder. By registering the images the field of view of the microscope can be widened, preserving the resolution while reconstructing the whole cell culture and permitting the microscopist to interactively explore the cell culture. In the second case study, the registration of terrestrial satellite images acquired by a camera integral with the satellite is utilized to estimate its three-dimensional orientation from visual data, for automatic guidance purposes. Critical aspects of these applications are emphasized and the choices adopted are motivated accordingly. Results are discussed in view of promising future developments.
Resumo:
Generic object recognition is an important function of the human visual system and everybody finds it highly useful in their everyday life. For an artificial vision system it is a really hard, complex and challenging task because instances of the same object category can generate very different images, depending of different variables such as illumination conditions, the pose of an object, the viewpoint of the camera, partial occlusions, and unrelated background clutter. The purpose of this thesis is to develop a system that is able to classify objects in 2D images based on the context, and identify to which category the object belongs to. Given an image, the system can classify it and decide the correct categorie of the object. Furthermore the objective of this thesis is also to test the performance and the precision of different supervised Machine Learning algorithms in this specific task of object image categorization. Through different experiments the implemented application reveals good categorization performances despite the difficulty of the problem. However this project is open to future improvement; it is possible to implement new algorithms that has not been invented yet or using other techniques to extract features to make the system more reliable. This application can be installed inside an embedded system and after trained (performed outside the system), so it can become able to classify objects in a real-time. The information given from a 3D stereocamera, developed inside the department of Computer Engineering of the University of Bologna, can be used to improve the accuracy of the classification task. The idea is to segment a single object in a scene using the depth given from a stereocamera and in this way make the classification more accurate.
Resumo:
A method for automatic scaling of oblique ionograms has been introduced. This method also provides a rejection procedure for ionograms that are considered to lack sufficient information, depicting a very good success rate. Observing the Kp index of each autoscaled ionogram, can be noticed that the behavior of the autoscaling program does not depend on geomagnetic conditions. The comparison between the values of the MUF provided by the presented software and those obtained by an experienced operator indicate that the procedure developed for detecting the nose of oblique ionogram traces is sufficiently efficient and becomes much more efficient as the quality of the ionograms improves. These results demonstrate the program allows the real-time evaluation of MUF values associated with a particular radio link through an oblique radio sounding. The automatic recognition of a part of the trace allows determine for certain frequencies, the time taken by the radio wave to travel the path between the transmitter and receiver. The reconstruction of the ionogram traces, suggests the possibility of estimating the electron density between the transmitter and the receiver, from an oblique ionogram. The showed results have been obtained with a ray-tracing procedure based on the integration of the eikonal equation and using an analytical ionospheric model with free parameters. This indicates the possibility of applying an adaptive model and a ray-tracing algorithm to estimate the electron density in the ionosphere between the transmitter and the receiver An additional study has been conducted on a high quality ionospheric soundings data set and another algorithm has been designed for the conversion of an oblique ionogram into a vertical one, using Martyn's theorem. This allows a further analysis of oblique soundings, throw the use of the INGV Autoscala program for the automatic scaling of vertical ionograms.
Resumo:
In recent years, Deep Learning techniques have shown to perform well on a large variety of problems both in Computer Vision and Natural Language Processing, reaching and often surpassing the state of the art on many tasks. The rise of deep learning is also revolutionizing the entire field of Machine Learning and Pattern Recognition pushing forward the concepts of automatic feature extraction and unsupervised learning in general. However, despite the strong success both in science and business, deep learning has its own limitations. It is often questioned if such techniques are only some kind of brute-force statistical approaches and if they can only work in the context of High Performance Computing with tons of data. Another important question is whether they are really biologically inspired, as claimed in certain cases, and if they can scale well in terms of "intelligence". The dissertation is focused on trying to answer these key questions in the context of Computer Vision and, in particular, Object Recognition, a task that has been heavily revolutionized by recent advances in the field. Practically speaking, these answers are based on an exhaustive comparison between two, very different, deep learning techniques on the aforementioned task: Convolutional Neural Network (CNN) and Hierarchical Temporal memory (HTM). They stand for two different approaches and points of view within the big hat of deep learning and are the best choices to understand and point out strengths and weaknesses of each of them. CNN is considered one of the most classic and powerful supervised methods used today in machine learning and pattern recognition, especially in object recognition. CNNs are well received and accepted by the scientific community and are already deployed in large corporation like Google and Facebook for solving face recognition and image auto-tagging problems. HTM, on the other hand, is known as a new emerging paradigm and a new meanly-unsupervised method, that is more biologically inspired. It tries to gain more insights from the computational neuroscience community in order to incorporate concepts like time, context and attention during the learning process which are typical of the human brain. In the end, the thesis is supposed to prove that in certain cases, with a lower quantity of data, HTM can outperform CNN.
Resumo:
We present a new method for the enhancement of speech. The method is designed for scenarios in which targeted speaker enrollment as well as system training within the typical noise environment are feasible. The proposed procedure is fundamentally different from most conventional and state-of-the-art denoising approaches. Instead of filtering a distorted signal we are resynthesizing a new “clean” signal based on its likely characteristics. These characteristics are estimated from the distorted signal. A successful implementation of the proposed method is presented. Experiments were performed in a scenario with roughly one hour of clean speech training data. Our results show that the proposed method compares very favorably to other state-of-the-art systems in both objective and subjective speech quality assessments. Potential applications for the proposed method include jet cockpit communication systems and offline methods for the restoration of audio recordings.
Resumo:
We present a new approach for corpus-based speech enhancement that significantly improves over a method published by Xiao and Nickel in 2010. Corpus-based enhancement systems do not merely filter an incoming noisy signal, but resynthesize its speech content via an inventory of pre-recorded clean signals. The goal of the procedure is to perceptually improve the sound of speech signals in background noise. The proposed new method modifies Xiao's method in four significant ways. Firstly, it employs a Gaussian mixture model (GMM) instead of a vector quantizer in the phoneme recognition front-end. Secondly, the state decoding of the recognition stage is supported with an uncertainty modeling technique. With the GMM and the uncertainty modeling it is possible to eliminate the need for noise dependent system training. Thirdly, the post-processing of the original method via sinusoidal modeling is replaced with a powerful cepstral smoothing operation. And lastly, due to the improvements of these modifications, it is possible to extend the operational bandwidth of the procedure from 4 kHz to 8 kHz. The performance of the proposed method was evaluated across different noise types and different signal-to-noise ratios. The new method was able to significantly outperform traditional methods, including the one by Xiao and Nickel, in terms of PESQ scores and other objective quality measures. Results of subjective CMOS tests over a smaller set of test samples support our claims.