Research in object detection and recognition in cluttered scenes requires large image collections with ground truth labels. The labels should provide information about the object classes present in each image, as well as their shape and locations, and possibly other attributes such as pose. Such data is useful for testing, as well as for supervised learning. This project provides a web-based annotation tool that makes it easy to annotate images, and to instantly sharesuch annotations with the community. This tool, plus an initial set of 10,000 images (3000 of which have been labeled), can be found at http://www.csail.mit.edu/$\sim$brussell/research/LabelMe/intro.html

Este estudo se dá a partir de uma série de trabalhos em desenho, fotografia, instalação e livro de artista, desenvolvidos entre 2012 e 2014, tendo como eixo um arquivo de desenhos realizados em situações cotidianas. Interessa aqui pensar a prática constante do desenho como ação cotidiana em pequenos suportes (folhas avulsas, cadernos de formatos modestos), sua acumulação através dos anos, sua utilização como extensão física da memória. O trabalho acontece na construção de novas totalidades a partir de fragmentos que compõem recortes transversais no tempo, em uma série de propostas para dar a ver esses desenhos, no diálogo sugerido pelo encontro de imagens realizadas em momentos distantes entre si ou no encontro destas imagens com o espaço da prática, assim como nas fricções e ambiguidades entre os desenhos de esboço, da ordem do privado, e trabalhos finalizados de ordem pública. / The present study’s starting point is a series of works in drawing, photography, installation and artist's book, which have been developed from 2012 to 2014, having as na axis the archive of drawings made in everyday situations. The focus here is to think about the constant practice of drawing as a daily action on small structures (single sheets, modest format notebooks), its accumulation over the years, its use as an extension of physical memory. The work takes place in the construction of new wholes from fragments through transverse cuts in time, in a series of proposals for seeing these drawings, the suggested dialogue between images made at distinct moments or in relation to the atelier, and the friction between the sketches, that are private, and finished works, which are public.

Este estudo se dá a partir de uma série de trabalhos em desenho, fotografia, instalação e livro de artista, desenvolvidos entre 2012 e 2014, tendo como eixo um arquivo de desenhos realizados em situações cotidianas. Interessa aqui pensar a prática constante do desenho como ação cotidiana em pequenos suportes (folhas avulsas, cadernos de formatos modestos), sua acumulação através dos anos, sua utilização como extensão física da memória. O trabalho acontece na construção de novas totalidades a partir de fragmentos que compõem recortes transversais no tempo, em uma série de propostas para dar a ver esses desenhos, no diálogo sugerido pelo encontro de imagens realizadas em momentos distantes entre si ou no encontro destas imagens com o espaço da prática, assim como nas fricções e ambiguidades entre os desenhos de esboço, da ordem do privado, e trabalhos finalizados de ordem pública. / The present study’s starting point is a series of works in drawing, photography, installation and artist's book, which have been developed from 2012 to 2014, having as na axis the archive of drawings made in everyday situations. The focus here is to think about the constant practice of drawing as a daily action on small structures (single sheets, modest format notebooks), its accumulation over the years, its use as an extension of physical memory. The work takes place in the construction of new wholes from fragments through transverse cuts in time, in a series of proposals for seeing these drawings, the suggested dialogue between images made at distinct moments or in relation to the atelier, and the friction between the sketches, that are private, and finished works, which are public.

The sweetpotato whitefly Bemisia tabaci is a highly destructive agricultural and ornamental crop pest. It damages host plants through both phloem feeding and vectoring plant pathogens. Introductions of B. tabaci are difficult to quarantine and eradicate because of its high reproductive rates, broad host plant range, and insecticide resistance. A total of 791 Gb of raw DNA sequence from whole genome shotgun sequencing, and 13 BAC pooling libraries were generated by Illumina sequencing using different combinations of mate-pair and pair-end libraries. Assembly gave a final genome with a scaffold N50 of 437 kb, and a total length of 658 Mb. Annotation of repetitive elements and coding regions resulted in 265.0 Mb TEs (40.3%) and 20 786 protein-coding genes with putative gene family expansions, respectively. Phylogenetic analysis based on orthologs across 14 arthropod taxa suggested that MED/Q is clustered into a hemipteran clade containing A. pisum and is a sister lineage to a clade containing both R. prolixus and N. lugens. Genome completeness, as estimated using the CEGMA and Benchmarking Universal Single-Copy Orthologs pipelines, reached 96% and 79%. These MED/Q genomic resources lay a foundation for future 'pan-genomic' comparisons of invasive vs. noninvasive, invasive vs. invasive, and native vs. exotic Bemisia, which, in return, will open up new avenues of investigation into whitefly biology, evolution, and management.

L’imagerie médicale et en particulier l’Imagerie par Résonance Magnétique (IRM) occupe une place de choix dans les décisions médicales. Malgré des techniques et des pratiques d’examens comparables les industriels du secteur utilisent un vocabulaire différent pour décrire les événements de l’expérience IRM. Les ontologies permettent de résoudre cette problématique. En les intégrant dans un système informatique nous avons choisi de proposer des solutions innovantes pour trois situations quotidiennes : l’annotation d’examen, la reconnaissance et la correction d’artéfact et l’aide à la prescription d’examen. Les connaissances du domaine IRM sont issues de la littérature et de la pratique quotidienne. DICOM, élément incontournable à l’échange de données en imagerie médicale, a été le point de départ de l’élaboration de l’ontologie. Les connaissances sur les artéfacts en IRM sont en partie issues d’une collaboration avec l’université de Texas A&M Temple. Pour l’aide à la prescription d’examen nous avons choisi une situation clinique représentative la demande d’IRM dans le cadre des traumatismes du genou. Les statistiques proviennent d’une étude réalisée dans le service de médecine du sport du CHU de Rennes. Les systèmes élaborés permettent aux utilisateurs d’utiliser les ontologies sans y être confronté et permettent l’analyse de l’entête DICOM d’une image, l’annotation de l’image, la comparaison d’une image IRM pour la correction d’artéfact et l’aide à la prescription d’IRM est sous la forme d’un serveur web permettant à l’utilisateur de connaître en fonction des signes cliniques présents, la probabilité d’avoir une lésion lors de la réalisation de l’IRM. Nous avons démontré la possibilité d’utilisation des ontologies pour améliorer l’exercice quotidien des praticiens. Les techniques utilisées montre la possibilité d’utiliser les ontologies en les associant aux images et aux probabilités. Le système choisi, interface permettra l’évolution vers une technologie de type webservice. / Magnetic resonance imaging (MRI) is a key examination in medical decision making. Despite MRI technics are slightly similar, each industrial has developed his own vocabulary to describe the MRI experience. Ontologies have been developed to help in such situations. We have decided to create IT solutions using ontology for three daily radiological situations: exams annotation, MRI artifacts recognition and correction and exam appropriateness. The domain knowledge is extracted from literature and everyday practice. DICOM, as key element for data exchange in radiology, has been used to create the ontology.  Concerning MRI artifacts, a part of the knowledge comes from a collaborative work with the university of Texas A&M Temple. Concerning exam appropriateness, we have chosen a representative clinical situation: interest of knee MRI in case of knee trauma. The statistical data are coming from a clinical study done in the CHU of Rennes. Our systems allow users to take benefits of ontology without facing it. They give a DICOM header analysis, proposed an image annotation, compare image to correct MRI artifacts and help physicians to judge MRI appropriateness in case of knee trauma. We have demonstrated that ontologies could be used to improve daily practice in radiology and that ontologies could be associated to image and statistical data. Future of this work could be a system transformation into a web service.

À l’heure actuelle, l’Internet est devenu une des sources d’information les plus importantes dans des nombreux domaines, dont le domaine de la santé. Les images médicales portent des informations importantes pour le diagnostic, l’enseignement et la recherche, des informations parfois difficiles à décrire textuellement. Avec le développement des technologies d’acquisition, les images médicales sont de plus en plus nombreuses dans les bases d’images et dans les documents en ligne. Vu l’importance des images médicales pour le diagnostic, l’enseignement et même l’éducation civique, des méthodes et des outils efficaces d’annotation, d’indexation et de recherche des images médicales sont nécessaires. Dans ce contexte, nous proposons une architecture pour l’annotation des images médicales incluses dans des documents de santé en ligne. Notre système extrait des informations médicales spécifiques (i.e. modalité médicale, région anatomique) à partir de plusieurs sources, et combine ces informations pour annoter les images. Cette annotation est nécessaire pour pouvoir retrouver les images à l’intérieur des documents sur le web. Dans nos expérimentations, nous avons implémenté deux approches (chacune basée sur une source d’information) : une première orientée vers le contenu des images, et une deuxième orientée sur le contexte des images (régions textuelles associées aux images). / Today, Internet has become a major source of information in many areas, including health. Medical images are carrying crucial information for diagnostic, teaching and research, moste of the time this information being very difficult to describe using only text. In this context, we are proposing architecture for the annotation of medical images included in online health documents. Our system extracts specific medical information (modality, anatomical regions ...) from several sources and combines this information to annotate the images. This annotation is necessary to be able to search for images inside documents. We implemented two distinct approaches (each based on a different type of information) : one oriented towards the image content, and a second oriented on the image context (image-related textual regions).

Highlighting is an annotation method frequently used to indicate importance. This document introduces context highlighting of web documents and demonstrates that passive readers may benefit from summaries produced by context/keyword highlighting. HighBrow, a prototype browser developed to enable active readers to create context/keyword summaries from web documents, is introduced. This document provides the results of an experiment using three groups of passive readers: the first group reading an entire document; the second group reading a context/keyword summary provided by HighBrow; and the third group reading a keyword only summary (provided by a modified version of HighBrow). The experiment was developed to measure quiz performance, preparation time, and efficiency (quiz score divided by time).

In order to understand the regulation of gene expression in organisms, precise genome annotation is essential. In recent years, RNA-Seq has become a potent method for generating and improving genome annotations. However, this Approach is time consuming and often inconsistently performed when done manually. In particular, the discovery of non-coding RNAs benefits strongly from the application of RNA-Seq data but requires significant amounts of expert knowledge and is labor-intensive. As a part of my doctoral study, I developed a modular tool called ANNOgesic that can detect numerous transcribed genomic features, including non-coding RNAs, based on RNA-Seq data in a precise and automatic fashion with a focus on bacterial and achaeal species. The software performs numerous analyses and generates several visualizations. It can generate annotations of high-Resolution that are hard to produce using traditional annotation tools that are based only on genome sequences. ANNOgesic can detect numerous novel genomic Features like UTR-derived small non-coding RNAs for which no other tool has been developed before. ANNOgesic is available under an open source license (ISCL) at https://github.com/Sung-Huan/ANNOgesic. My doctoral work not only includes the development of ANNOgesic but also its application to annotate the transcriptome of Staphylococcus aureus HG003 - a strain which has been a insightful model in infection biology. Despite its potential as a model, a complete genome sequence and annotations have been lacking for HG003. In order to fill this gap, the annotations of this strain, including sRNAs and their functions, were generated using ANNOgesic by analyzing differential RNA-Seq data from 14 different samples (two media conditions with seven time points), as well as RNA-Seq data generated after transcript fragmentation. ANNOgesic was also applied to annotate several bacterial and archaeal genomes, and as part of this its high performance was demonstrated. In summary, ANNOgesic is a powerful computational tool for RNA-Seq based annotations and has been successfully applied to several species. / Exakte Genomannotationen sind essentiell für das Verständnis Genexpressionsregulation in verschiedenen Organismen. In den letzten Jahren entwickelte sich RNA-Seq zu einer äußerst wirksamen Methode, um solche Genomannotationen zu erstellen und zu verbessern. Allerdings ist das Erstellen von Genomannotationen bei manueller Durchführung noch immer ein zeitaufwändiger und inkonsistenter Prozess. Die Verwendung von RNA-Seq-Daten begünstigt besonders die Identifizierung von nichtkodierenden RNAs, was allerdings arbeitsintensiv ist und fundiertes Expertenwissen erfordert. Ein Teil meiner Promotion bestand aus der Entwicklung eines modularen Tools namens ANNOgesic, das basierend auf RNA-Seq-Daten in der Lage ist, eine Vielzahl von Genombestandteilen, einschließlich nicht-kodierender RNAs, automatisch und präzise zu ermitteln. Das Hauptaugenmerk lag dabei auf der Anwendbarkeit für bakterielle und archaeale Genome. Die Software führt eine Vielzahl von Analysen durch und stellt die verschiedenen Ergebnisse grafisch dar. Sie generiert hochpräzise Annotationen, die nicht unter Verwendung herkömmlicher Annotations-Tools auf Basis von Genomsequenzen erzeugt werden könnten. Es kann eine Vielzahl neuer Genombestandteile, wie kleine nicht-kodierende RNAs in UTRs, ermitteln, welche von bisherigen Programme nicht vorhergesagt werden können. ANNOgesic ist unter einer Open-Source-Lizenz (ISCL) auf https://github.com/Sung-Huan/ANNOgesic verfügbar. Meine Forschungsarbeit beinhaltet nicht nur die Entwicklung von ANNOgesic, sondern auch dessen Anwendung um das Transkriptom des Staphylococcus aureus-Stamms HG003 zu annotieren. Dieser ist einem Derivat von S. aureus NCTC8325 - ein Stamm, Dear ein bedeutendes Modell in der Infektionsbiologie darstellt. Zum Beispiel wurde er für die Untersuchung von Antibiotikaresistenzen genutzt, da er anfällig für alle bekannten Antibiotika ist. Der Elternstamm NCTC8325 besitzt zwei Mutationen im regulatorischen Genen (rsbU und tcaR), die Veränderungen der Virulenz zur Folge haben und die in Stamm HG003 auf die Wildtypsequenz zurückmutiert wurden. Dadurch besitzt S. aureus HG003 das vollständige, ursprüngliche Regulationsnetzwerk und stellt deshalb ein besseres Modell zur Untersuchung von sowohl Virulenz als auch Antibiotikaresistenz dar. Trotz seines Modellcharakters fehlten für HG003 bisher eine vollständige Genomsequenz und deren Annotationen. Um diese Lücke zu schließen habe ich als Teil meiner Promotion mit Hilfe von ANNOgesic Annotationen für diesen Stamm, einschließlich sRNAs und ihrer Funktionen, generiert. Dafür habe ich Differential RNA-Seq-Daten von 14 verschiedenen Proben (zwei Mediumsbedingungen mit sieben Zeitpunkten) sowie RNA-Seq-Daten, die von fragmentierten Transkripten generiert wurden, analysiert. Neben S. aureus HG003 wurde ANNOgesic auf eine Vielzahl von Bakterien- und Archaeengenome angewendet und dabei wurde eine hohe Performanz demonstriert. Zusammenfassend kann gesagt werden, dass ANNOgesic ein mächtiges bioinformatisches Werkzeug für die RNA-Seq-basierte Annotationen ist und für verschiedene Spezies erfolgreich angewandt wurde.

Haloxylon persicum is a xerophytic desert tree that grows mostly in deserts in West and Central Asia. This tree is very tolerant to the harsh conditions of deserts, mainly drought and heat. As a part of the Kingdom of Saudi Arabia Native Genome Project, a voucher specimen was identified, and the genome of this plant was sequenced, assembled, and annotated. The chromosome level assembly was performed using the integration of PacBio Hifi reads and Bionano optical maps, resulting in 9 chromosome-sized molecules that only exhibit 3 gaps located in highly repetitive regions. The annotation of the transposable elements in the genome shows that more than 55% of the genome consists of transposable elements. Moreover, genes were predicted using Iso-seq and RNA-seq and annotated using publicly available protein databases, resulting in the identification of more than 45,000 predicted genes, of which ≈ 10,000 have RNA evidence. The genome assembly and annotation of Haloxylon persicum will: provide valuable insight on the evolutionary history of desert plants, aid in discovering the mechanisms developed by this species to cope with the extreme desert conditions and unveil the possibilities and opportunities of neo-domesticating this plant. Furthermore, this assembly can serve as a reference for assembling other plant species in the KSA Native genome project or any other project worldwide.

The Genome Reverse Compiler (GRC) is an annotation tool for prokaryotic genomes. Its name and philosophy are based on analogy with a high-level programming language compiler. In this analogy, the genome is a program in a certain low-level language that humans cannot understand. Given the sequence of any prokaryotic genome, GRC produces its corresponding "high-level program"--its annotation. GRC works in a completely automatic manner, using standard input and output formats. The goal is to provide an open-source, easy-to-run, very efficient annotation program. / Master of Science