Search results for "deep learning"

showing 10 items of 337 documents

Deep learning techniques for visual object tracking

2023

Visual object tracking plays a crucial role in various vision systems, including biometric analysis, medical imaging, smart traffic systems, and video surveillance. Despite notable advancements in visual object tracking over the past few decades, many tracking algorithms still face challenges due to factors like illumination changes, deformation, and scale variations. This thesis is divided into three parts. The first part introduces the visual object tracking problem and discusses the traditional approaches that have been used to study it. We then propose a novel method called Tracking by Iterative Multi-Refinements, which addresses the issue of locating the target by redefining the search…

Settore ING-INF/05 - Sistemi Di Elaborazione Delle InformazioniDeep LearningVisual Object TrackingFast LearningReinforcement LearningCNN
researchProduct

Towards a deep-learning-based methodology for supporting satire detection

2021

This paper describes an approach for supporting automatic satire detection through effective deep learning (DL) architecture that has been shown to be useful for addressing sarcasm/irony detection problems. We both trained and tested the system exploiting articles derived from two important satiric blogs, Lercio and IlFattoQuotidiano, and significant Italian newspapers.

Settore ING-INF/05 - Sistemi Di Elaborazione Delle InformazioniDeep Neural NetworksSettore INF/01 - InformaticaVisual languagesNatural Language processingDeep learningSatire Detection
researchProduct

Challenges in Image Matching for Cultural Heritage: An Overview and Perspective

2022

Image matching, as the task of finding correspondences in images, is the upstream component of vision and photogrammetric applications aiming at the reconstruction of 3D scenes, their understanding and comparison. Such applications are of special importance in the context of cultural heritage, as they can support archaeologists to digitally preserve, restore and analyze antiquities, but also to compare their changes over time. The success of deep learning, now firmly established, paired with the evolution of computer hardware, has led to many advances in image processing, including image matching. Despite this progress, image matching still offers challenges, in terms of the matching proces…

Settore ING-INF/05 - Sistemi Di Elaborazione Delle InformazioniImage matchingSettore INF/01 - InformaticaSfMSIFTCultural heritageDeep learning
researchProduct

McRock at SemEval-2022 Task 4: Patronizing and Condescending Language Detection using Multi-Channel CNN, Hybrid LSTM, DistilBERT and XLNet

2022

In this paper we propose four deep learning models for the task of detecting and classifying Patronizing and Condescending Language (PCL) using a corpus of over 13,000 annotated paragraphs in English. The task, hosted at SemEval-2022, consists of two different subtasks. The Subtask 1 is a binary classification problem. Namely, given a paragraph, a system must predict whether or not it contains any form of PCL. The Subtask 2 is a multi-label classification task. Given a paragraph, a system must identify which PCL categories express the condescension. A paragraph might contain one or more categories of PCL. To face with the first subtask we propose a multi-channel Convolutional Neural Network…

Settore ING-INF/05 - Sistemi Di Elaborazione Delle InformazioniNLP Deep Learning Machine Learning XLNet CNN DistilBERT PCLProceedings of the 16th International Workshop on Semantic Evaluation (SemEval-2022)
researchProduct

PHOTOGRAMMETRY NOW AND THEN - FROM HAND-CRAFTED TO DEEP-LEARNING TIE POINTS

2022

Abstract. Historical images provide a valuable source of information exploited by several kinds of applications, such as the monitoring of cities and territories, the reconstruction of destroyed buildings, and are increasingly being shared for cultural promotion projects through virtual reality or augmented reality applications. Finding reliable and accurate matches between historical and present images is a fundamental step for such tasks since they require to co-register the present 3D scene with the past one. Classical image matching solutions are sensitive to strong radiometric variations within the images, which are particularly relevant in these multi-temporal contexts due to differen…

Settore ING-INF/05 - Sistemi Di Elaborazione Delle InformazioniRANSACtie pointsSettore INF/01 - Informaticadeep learningimage matchingcultural heritageHistorical imageslocal features
researchProduct

A Novel Approach for Supporting Italian Satire Detection Through Deep Learning

2021

Satire is a way of criticizing people (or ideas) by ridiculing them on political, social, and morals topics often used to denounce political and societal problems, leveraging comedic devices such as parody exaggeration, incongruity, etc.etera. Detecting satire is one of the most challenging computational linguistics tasks, natural language processing, and social multimedia sentiment analysis. In particular, as satirical texts include figurative communication for expressing ideas/opinions concerning people, sentiment analysis systems may be negatively affected; therefore, satire should be adequately addressed to avoid such systems’ performance degradation. This paper tackles automatic satire…

Settore ING-INF/05 - Sistemi Di Elaborazione Delle InformazioniSettore INF/01 - InformaticaSarcasmComputer scienceNatural language processingmedia_common.quotation_subjectSentiment analysisSatire detectionDeep learningContext (language use)Literal and figurative languageLinguisticsNewspaperPoliticsExaggerationComputational linguisticsmedia_common
researchProduct

Retinal image synthesis through the least action principle

2020

Eye fundus image analysis is a fundamental approach in medical diagnosis and follow-up ophthalmic diagnostics. Manual annotation by experts needs hard work, thus only a small set of annotated vessel structures is available. Examples such as DRIVE and STARE include small sets for training images of fundus image benchmarks. Moreover, there is no vessel structure annotation for a number of fundus image datasets. Synthetic images have been generated by using appropriate parameters for the modeling of vascular networks or by methods developing deep learning techniques and supported by performance hardware. Our methodology aims to produce high-resolution synthetic fundus images alternative to the…

Settore ING-INF/05 - Sistemi Di Elaborazione Delle InformazioniSettore INF/01 - Informaticapredictive evaluation diseasesComputer sciencebusiness.industryDeep learningComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISIONFundus (eye)Real imageSmall setPrinciple of least actionImage (mathematics)fundus image analysisAnnotationComputer visionArtificial intelligenceMedical diagnosisbusinessstatistical featuressynthetic retinal imagedata augmentation2020 5th International Conference on Intelligent Informatics and Biomedical Sciences (ICIIBMS)
researchProduct

Deep Metric Learning for Histopathological Image Classification

2022

Neural networks demonstrated to be effective in multiple classification tasks with performances that are similar to human capabilities. Notwithstanding, the viability of the application of this kind of tool in real cases passes through the possibility to interpret the provided results and let the human operator take his decision according to the information that is provided. This aspect is much more evident when the field of application is bound to people's health as for biomed-ical image classification. We propose for the classification of histopathological images a convolutional neural network that, through metric learning, learns a representation that gathers in homogeneous clusters the …

Settore ING-INF/05 - Sistemi Di Elaborazione Delle InformazioniembeddingSettore INF/01 - Informaticametric learningdeep learninghistopathological images2022 IEEE Eighth International Conference on Multimedia Big Data (BigMM)
researchProduct

Fake News Spreaders Detection: Sometimes Attention Is Not All You Need

2022

Guided by a corpus linguistics approach, in this article we present a comparative evaluation of State-of-the-Art (SotA) models, with a special focus on Transformers, to address the task of Fake News Spreaders (i.e., users that share Fake News) detection. First, we explore the reference multilingual dataset for the considered task, exploiting corpus linguistics techniques, such as chi-square test, keywords and Word Sketch. Second, we perform experiments on several models for Natural Language Processing. Third, we perform a comparative evaluation using the most recent Transformer-based models (RoBERTa, DistilBERT, BERT, XLNet, ELECTRA, Longformer) and other deep and non-deep SotA models (CNN,…

Settore ING-INF/05 - Sistemi Di Elaborazione Delle Informazionitext classificationcorpus linguisticSettore ING-INF/03 - Telecomunicazionifake newTwitterauthor profilingconvolutional neural networkdeep learningNatural Language Processing (NLP)user classificationfake news; misinformation; Natural Language Processing (NLP); transformers; Twitter; convolutional neural networks; text classification; deep learning; machine learning; user classification; author profiling; corpus linguistics; linguistic analysismachine learningtransformermisinformationlinguistic analysisInformation Systems
researchProduct

Toward a Collective Agenda on AI for Earth Science Data Analysis

2021

In the last years we have witnessed the fields of geosciences and remote sensing and artificial intelligence to become closer. Thanks to both the massive availability of observational data, improved simulations, and algorithmic advances, these disciplines have found common objectives and challenges to advance the modeling and understanding of the Earth system. Despite such great opportunities, we also observed a worrying tendency to remain in disciplinary comfort zones applying recent advances from artificial intelligence on well resolved remote sensing problems. Here we take a position on research directions where we think the interface between these fields will have the most impact and be…

Signal Processing (eess.SP)FOS: Computer and information sciences010504 meteorology & atmospheric sciencesGeneral Computer Science530 PhysicsInterface (Java)Computer Vision and Pattern Recognition (cs.CV)Earth sciencedata analysisComputer Science - Computer Vision and Pattern Recognition0211 other engineering and technologiesearth observation02 engineering and technology01 natural sciencesEnvironmental scienceData modelingFOS: Electrical engineering electronic engineering information engineeringClimate science1700 General Computer ScienceElectrical Engineering and Systems Science - Signal ProcessingElectrical and Electronic EngineeringInstrumentation021101 geological & geomatics engineering0105 earth and related environmental sciences11476 Digital Society Initiative3105 Instrumentation2208 Electrical and Electronic Engineering1900 General Earth and Planetary SciencesDeep learninginterpretable AIRemote sensingartificial intelligencehybrid modelsEarth system scienceAIRemote sensing (archaeology)10231 Institute for Computational ScienceGeneral Earth and Planetary SciencesPotential gameDisciplineIEEE Geoscience and Remote Sensing Magazine
researchProduct