972 resultados para Automatic Image Annotation
Resumo:
This paper proposes to enrich RBMTdictionaries with Named Entities(NEs) automatically acquired fromWikipedia. The method is appliedto the Apertium English-Spanishsystem and its performance comparedto that of Apertium with and withouthandtagged NEs. The system withautomatic NEs outperforms the onewithout NEs, while results vary whencompared to a system with handtaggedNEs (results are comparable forSpanish to English but slightly worstfor English to Spanish). Apart fromthat, adding automatic NEs contributesto decreasing the amount of unknownterms by more than 10%.
Resumo:
The paper deals with the development and application of the methodology for automatic mapping of pollution/contamination data. General Regression Neural Network (GRNN) is considered in detail and is proposed as an efficient tool to solve this problem. The automatic tuning of isotropic and an anisotropic GRNN model using cross-validation procedure is presented. Results are compared with k-nearest-neighbours interpolation algorithm using independent validation data set. Quality of mapping is controlled by the analysis of raw data and the residuals using variography. Maps of probabilities of exceeding a given decision level and ?thick? isoline visualization of the uncertainties are presented as examples of decision-oriented mapping. Real case study is based on mapping of radioactively contaminated territories.
Resumo:
BACKGROUND: The GENCODE consortium was formed to identify and map all protein-coding genes within the ENCODE regions. This was achieved by a combination of initial manual annotation by the HAVANA team, experimental validation by the GENCODE consortium and a refinement of the annotation based on these experimental results. RESULTS: The GENCODE gene features are divided into eight different categories of which only the first two (known and novel coding sequence) are confidently predicted to be protein-coding genes. 5' rapid amplification of cDNA ends (RACE) and RT-PCR were used to experimentally verify the initial annotation. Of the 420 coding loci tested, 229 RACE products have been sequenced. They supported 5' extensions of 30 loci and new splice variants in 50 loci. In addition, 46 loci without evidence for a coding sequence were validated, consisting of 31 novel and 15 putative transcripts. We assessed the comprehensiveness of the GENCODE annotation by attempting to validate all the predicted exon boundaries outside the GENCODE annotation. Out of 1,215 tested in a subset of the ENCODE regions, 14 novel exon pairs were validated, only two of them in intergenic regions. CONCLUSION: In total, 487 loci, of which 434 are coding, have been annotated as part of the GENCODE reference set available from the UCSC browser. Comparison of GENCODE annotation with RefSeq and ENSEMBL show only 40% of GENCODE exons are contained within the two sets, which is a reflection of the high number of alternative splice forms with unique exons annotated. Over 50% of coding loci have been experimentally verified by 5' RACE for EGASP and the GENCODE collaboration is continuing to refine its annotation of 1% human genome with the aid of experimental validation.
Resumo:
Background: We examined one's own body image perception and its association with reported weight-related behavior among adolescents of a rapidly developing country in the African region. Methods: We conducted a school-based survey of 1432 students aged 11-17 years in the Seychelles. Weight and height were measured, and thinness, normal weight and overweight were assessed along standard criteria. A self-administered and anonymous questionnaire was administered. Perception of body image was assessed using both a closed-ended question (CEQ) and the Stunkard's pictorial silhouettes (SPS). Finally, a question assessed voluntary attempts to change weight. Results: Overall, 14.1% of the students were thin, 63.9% were normal-weight, and 22.0% were overweight or obese. There was fair agreement between actual weight status and self-perceived body image based on either CEQ or SPS. However, a substantial proportion of the overweight students did not consider themselves as overweight (SPS: 24%, CEQ: 34%) and, inversely, a substantial proportion of the normal-weight students considered themselves as too thin (SPS: 29%, CEQ: 15%). Among the overweight students, an adequate attempt to lose weight was reported more often by boys and girls who perceived themselves as overweight vs. not overweight (72-88% vs. 40-71%, p <0.05 for most comparisons). Among the normal-weight students, an inadequate attempt to gain weight was reported more often by boys and girls who perceived themselves as thin vs. not thin (27-68% vs. 11-19%, p <0.05). Girls had leaner own body ideals than boys. Conclusions: We found that substantial proportions of overweight students did not perceive themselves as overweight and/or did not want to lose weight and, inversely, that many normalweight students perceived themselves as too thin and/or wanted to gain weight: this points to forces that can drive the upwards overweight trends. Appropriate perception of one's weight was associated with adequate weight-control behavior, although not strongly, emphasizing that appropriate weight perception is only one of several factors driving adequate weight-related behavior. These findings emphasize the need to address appropriate perception of one's own weight and adequate weight-related behavior in adolescents for both individual and community weight-related interventions.
Resumo:
RESUME Les fibres textiles sont des produits de masse utilisés dans la fabrication de nombreux objets de notre quotidien. Le transfert de fibres lors d'une action délictueuse est dès lors extrêmement courant. Du fait de leur omniprésence dans notre environnement, il est capital que l'expert forensique évalue la valeur de l'indice fibres. L'interprétation de l'indice fibres passe par la connaissance d'un certain nombre de paramètres, comme la rareté des fibres, la probabilité de leur présence par hasard sur un certain support, ainsi que les mécanismes de transfert et de persistance des fibres. Les lacunes les plus importantes concernent les mécanismes de transfert des fibres. A ce jour, les nombreux auteurs qui se sont penchés sur le transfert de fibres ne sont pas parvenus à créer un modèle permettant de prédire le nombre de fibres que l'on s'attend à retrouver dans des circonstances de contact données, en fonction des différents paramètres caractérisant ce contact et les textiles mis en jeu. Le but principal de cette recherche est de démontrer que la création d'un modèle prédictif du nombre de fibres transférées lors d'un contact donné est possible. Dans le cadre de ce travail, le cas particulier du transfert de fibres d'un tricot en laine ou en acrylique d'un conducteur vers le dossier du siège de son véhicule a été étudié. Plusieurs caractéristiques des textiles mis en jeu lors de ces expériences ont été mesurées. Des outils statistiques (régression linéaire multiple) ont ensuite été utilisés sur ces données afin d'évaluer l'influence des caractéristiques des textiles donneurs sur le nombre de fibres transférées et d'élaborer un modèle permettant de prédire le nombre de fibres qui vont être transférées à l'aide des caractéristiques influençant significativement le transfert. Afin de faciliter la recherche et le comptage des fibres transférées lors des expériences de transfert, un appareil de recherche automatique des fibres (liber finder) a été utilisé dans le cadre de cette recherche. Les tests d'évaluation de l'efficacité de cet appareil pour la recherche de fibres montrent que la recherche automatique est globalement aussi efficace qu'une recherche visuelle pour les fibres fortement colorées. Par contre la recherche automatique perd de son efficacité pour les fibres très pâles ou très foncées. Une des caractéristiques des textiles donneurs à étudier est la longueur des fibres. Afin de pouvoir évaluer ce paramètre, une séquence d'algorithmes de traitement d'image a été implémentée. Cet outil permet la mesure de la longueur d'une fibre à partir de son image numérique à haute résolution (2'540 dpi). Les tests effectués montrent que les mesures ainsi obtenues présentent une erreur de l'ordre du dixième de millimètre, ce qui est largement suffisant pour son utilisation dans le cadre de cette recherche. Les résultats obtenus suite au traitement statistique des résultats des expériences de transfert ont permis d'aboutir à une modélisation du phénomène du transfert. Deux paramètres sont retenus dans le modèle: l'état de la surface du tissu donneur et la longueur des fibres composant le tissu donneur. L'état de la surface du tissu est un paramètre tenant compte de la quantité de fibres qui se sont détachées de la structure du tissu ou qui sont encore faiblement rattachées à celle-ci. En effet, ces fibres sont les premières à se transférer lors d'un contact, et plus la quantité de ces fibres par unité de surface est importante, plus le nombre de fibres transférées sera élevé. La longueur des fibres du tissu donneur est également un paramètre important : plus les fibres sont longues, mieux elles sont retenues dans la structure du tissu et moins elles se transféreront. SUMMARY Fibres are mass products used to produce numerous objects encountered everyday. The transfer of fibres during a criminal action is then very common. Because fibres are omnipresent in our environment, the forensic expert has to evaluate the value of the fibre evidence. To interpret fibre evidence, the expert has to know some parameters as frequency of fibres,' probability of finding extraneous fibres by chance on a given support, and transfer and persistence mechanisms. Fibre transfer is one of the most complex parameter. Many authors studied fibre transfer mechanisms but no model has been created to predict the number of fibres transferred expected in a given type of contact according to parameters as characteristics of the contact and characteristics of textiles. The main purpose of this research is to demonstrate that it is possible to create a model to predict the number of fibres transferred during a contact. In this work, the particular case of the transfer of fibres from a knitted textile in wool or in acrylic of a driver to the back of a carseat has been studied. Several characteristics of the textiles used for the experiments were measured. The data obtained were then treated with statistical tools (multiple linear regression) to evaluate the influence of the donor textile characteristics on the number of úbers transferred, and to create a model to predict this number of fibres transferred by an equation containing the characteristics having a significant influence on the transfer. To make easier the searching and the counting of fibres, an apparatus of automatic search. of fibers (fiber finder) was used. The tests realised to evaluate the efficiency of the fiber finder shows that the results obtained are generally as efficient as for visual search for well-coloured fibres. However, the efficiency of automatic search decreases for pales and dark fibres. One characteristic of the donor textile studied was the length of the fibres. To measure this parameter, a sequence of image processing algorithms was implemented. This tool allows to measure the length of a fibre from it high-resolution (2'540 dpi) numerical image. The tests done shows that the error of the measures obtained are about some tenths of millimetres. This precision is sufficient for this research. The statistical methods applied on the transfer experiment data allow to create a model of the transfer phenomenon. Two parameters are included in the model: the shedding capacity of the donor textile surface and the length of donor textile fibres. The shedding capacity of the donor textile surface is a parameter estimating the quantity of fibres that are not or slightly attached to the structure of the textile. These fibres are easily transferred during a contact, and the more this quantity of fibres is high, the more the number of fibres transferred during the contact is important. The length of fibres is also an important parameter: the more the fibres are long, the more they are attached in the structure of the textile and the less they are transferred during the contact.
Resumo:
Image filtering is a highly demanded approach of image enhancement in digital imaging systems design. It is widely used in television and camera design technologies to improve the quality of an output image to avoid various problems such as image blurring problem thatgains importance in design of displays of large sizes and design of digital cameras. This thesis proposes a new image filtering method basedon visual characteristics of human eye such as MTF. In contrast to the traditional filtering methods based on human visual characteristics this thesis takes into account the anisotropy of the human eye vision. The proposed method is based on laboratory measurements of the human eye MTF and takes into account degradation of the image by the latter. This method improves an image in the way it will be degraded by human eye MTF to give perception of the original image quality. This thesis gives a basic understanding of an image filtering approach and the concept of MTF and describes an algorithm to perform an image enhancement based on MTF of human eye. Performed experiments have shown quite good results according to human evaluation. Suggestions to improve the algorithm are also given for the future improvements.
Resumo:
The aim of the present study is to determine the level of correlation between the 3-dimensional (3D) characteristics of trabecular bone microarchitecture, as evaluated using microcomputed tomography (μCT) reconstruction, and trabecular bone score (TBS), as evaluated using 2D projection images directly derived from 3D μCT reconstruction (TBSμCT). Moreover, we have evaluated the effects of image degradation (resolution and noise) and X-ray energy of projection on these correlations. Thirty human cadaveric vertebrae were acquired on a microscanner at an isotropic resolution of 93μm. The 3D microarchitecture parameters were obtained using MicroView (GE Healthcare, Wauwatosa, MI). The 2D projections of these 3D models were generated using the Beer-Lambert law at different X-ray energies. Degradation of image resolution was simulated (from 93 to 1488μm). Relationships between 3D microarchitecture parameters and TBSμCT at different resolutions were evaluated using linear regression analysis. Significant correlations were observed between TBSμCT and 3D microarchitecture parameters, regardless of the resolution. Correlations were detected that were strongly to intermediately positive for connectivity density (0.711≤r(2)≤0.752) and trabecular number (0.584≤r(2)≤0.648) and negative for trabecular space (-0.407 ≤r(2)≤-0.491), up to a pixel size of 1023μm. In addition, TBSμCT values were strongly correlated between each other (0.77≤r(2)≤0.96). Study results show that the correlations between TBSμCT at 93μm and 3D microarchitecture parameters are weakly impacted by the degradation of image resolution and the presence of noise.
Resumo:
We study the impact of sampling theorems on the fidelity of sparse image reconstruction on the sphere. We discuss how a reduction in the number of samples required to represent all information content of a band-limited signal acts to improve the fidelity of sparse image reconstruction, through both the dimensionality and sparsity of signals. To demonstrate this result, we consider a simple inpainting problem on the sphere and consider images sparse in the magnitude of their gradient. We develop a framework for total variation inpainting on the sphere, including fast methods to render the inpainting problem computationally feasible at high resolution. Recently a new sampling theorem on the sphere was developed, reducing the required number of samples by a factor of two for equiangular sampling schemes. Through numerical simulations, we verify the enhanced fidelity of sparse image reconstruction due to the more efficient sampling of the sphere provided by the new sampling theorem.
Resumo:
In this paper, we consider active sampling to label pixels grouped with hierarchical clustering. The objective of the method is to match the data relationships discovered by the clustering algorithm with the user's desired class semantics. The first is represented as a complete tree to be pruned and the second is iteratively provided by the user. The active learning algorithm proposed searches the pruning of the tree that best matches the labels of the sampled points. By choosing the part of the tree to sample from according to current pruning's uncertainty, sampling is focused on most uncertain clusters. This way, large clusters for which the class membership is already fixed are no longer queried and sampling is focused on division of clusters showing mixed labels. The model is tested on a VHR image in a multiclass classification setting. The method clearly outperforms random sampling in a transductive setting, but cannot generalize to unseen data, since it aims at optimizing the classification of a given cluster structure.
Resumo:
Le partage et la réutilisation d'objets d'apprentissage est encore une utopie. La mise en commun de documents pédagogiques et leur adaptation à différents contextes ont fait l'objet de très nombreux travaux. L'un des aspects qui fait problème concerne leur description qui se doit d'être aussi précise que possible afin d'en faciliter la gestion et plus spécifiquement un accès ciblé. Cette description s'effectue généralement par l'instanciation d'un ensemble de descripteurs standardisés ou métadonnées (LOM, ARIADNE, DC, etc). Force est de constater que malgré l'existence de ces standards, dont certains sont relativement peu contraignants, peu de pédagogues ou d'auteurs se prêtent à cet exercice qui reste lourd et peu gratifiant. Nous sommes parti de l'idée que si l'indexation pouvait être réalisée automatiquement avec un bon degré d'exactitude, une partie de la solution serait trouvée. Pour ce, nous nous sommes tout d'abord penché sur l'analyse des facteurs bloquants de la génération manuelle effectuée par les ingénieurs pédagogiques de l'Université de Lausanne. La complexité de ces facteurs (humains et techniques) nous a conforté dans l'idée que la génération automatique de métadonnées était bien de nature à contourner les difficultés identifiées. Nous avons donc développé une application de génération automatique de métadonnées laquelle se focalise sur le contenu comme source unique d'extraction. Une analyse en profondeur des résultats obtenus, nous a permis de constater que : - Pour les documents non structurés : notre application présente des résultats satisfaisants en se basant sur les indicateurs de mesure de qualité des métadonnées (complétude, précision, consistance logique et cohérence). - Pour des documents structurés : la génération automatique s'est révélée peu satisfaisante dans la mesure où elle ne permet pas d'exploiter les éléments sémantiques (structure, annotations) qu'ils contiennent. Et dans ce cadre nous avons pensé qu'il était possible de faire mieux. C'est ainsi que nous avons poursuivi nos travaux afin de proposer une deuxième application tirant profit du potentiel des documents structurés et des langages de transformation (XSLT) qui s'y rapportent pour améliorer la recherche dans ces documents. Cette dernière exploite la totalité des éléments sémantiques (structure, annotations) et constitue une autre alternative à la recherche basée sur les métadonnées. De plus, la recherche basée sur les annotations et la structure offre comme avantage supplémentaire de permettre de retrouver, non seulement les documents eux-mêmes, mais aussi des parties de documents. Cette caractéristique apporte une amélioration considérable par rapport à la recherche par métadonnées qui ne donne accès qu'à des documents entiers. En conclusion nous montrerons, à travers des exemples appropriés, que selon le type de document : il est possible de procéder automatiquement à leur indexation pour faciliter la recherche de documents dès lors qu'il s'agit de documents non structurés ou d'exploiter directement leur contenu sémantique dès lors qu'il s'agit de documents structurés.
Resumo:
The topic of this thesis is studying how lesions in retina caused by diabetic retinopathy can be detected from color fundus images by using machine vision methods. Methods for equalizing uneven illumination in fundus images, detecting regions of poor image quality due toinadequate illumination, and recognizing abnormal lesions were developed duringthe work. The developed methods exploit mainly the color information and simpleshape features to detect lesions. In addition, a graphical tool for collecting lesion data was developed. The tool was used by an ophthalmologist who marked lesions in the images to help method development and evaluation. The tool is a general purpose one, and thus it is possible to reuse the tool in similar projects.The developed methods were tested with a separate test set of 128 color fundus images. From test results it was calculated how accurately methods classify abnormal funduses as abnormal (sensitivity) and healthy funduses as normal (specificity). The sensitivity values were 92% for hemorrhages, 73% for red small dots (microaneurysms and small hemorrhages), and 77% for exudates (hard and soft exudates). The specificity values were 75% for hemorrhages, 70% for red small dots, and 50% for exudates. Thus, the developed methods detected hemorrhages accurately and microaneurysms and exudates moderately.
Resumo:
This paper presents the evaluation results of the methods submitted to Challenge US: Biometric Measurements from Fetal Ultrasound Images, a segmentation challenge held at the IEEE International Symposium on Biomedical Imaging 2012. The challenge was set to compare and evaluate current fetal ultrasound image segmentation methods. It consisted of automatically segmenting fetal anatomical structures to measure standard obstetric biometric parameters, from 2D fetal ultrasound images taken on fetuses at different gestational ages (21 weeks, 28 weeks, and 33 weeks) and with varying image quality to reflect data encountered in real clinical environments. Four independent sub-challenges were proposed, according to the objects of interest measured in clinical practice: abdomen, head, femur, and whole fetus. Five teams participated in the head sub-challenge and two teams in the femur sub-challenge, including one team who tackled both. Nobody attempted the abdomen and whole fetus sub-challenges. The challenge goals were two-fold and the participants were asked to submit the segmentation results as well as the measurements derived from the segmented objects. Extensive quantitative (region-based, distance-based, and Bland-Altman measurements) and qualitative evaluation was performed to compare the results from a representative selection of current methods submitted to the challenge. Several experts (three for the head sub-challenge and two for the femur sub-challenge), with different degrees of expertise, manually delineated the objects of interest to define the ground truth used within the evaluation framework. For the head sub-challenge, several groups produced results that could be potentially used in clinical settings, with comparable performance to manual delineations. The femur sub-challenge had inferior performance to the head sub-challenge due to the fact that it is a harder segmentation problem and that the techniques presented relied more on the femur's appearance.
Resumo:
Tässä työssä raportoidaan hybridihitsauksesta otettujen suurnopeuskuvasarjojen automaattisen analyysijärjestelmän kehittäminen.Järjestelmän tarkoitus oli tuottaa tietoa, joka avustaisi analysoijaa arvioimaan kuvatun hitsausprosessin laatua. Tutkimus keskittyi valokaaren taajuuden säännöllisyyden ja lisäainepisaroiden lentosuuntien mittaamiseen. Valokaaria havaittiin kuvasarjoista sumean c-means-klusterointimenetelmän avullaja perättäisten valokaarien välistä aikaväliä käytettiin valokaaren taajuuden säännöllisyyden mittarina. Pisaroita paikannettiin menetelmällä, jossa yhdistyi pääkomponenttianalyysi ja tukivektoriluokitin. Kalman-suodinta käytettiin tuottamaan arvioita pisaroiden lentosuunnista ja nopeuksista. Lentosuunnanmääritysmenetelmä luokitteli pisarat niiden arvioitujen lentosuuntien perusteella. Järjestelmän kehittämiseen käytettävissä olleet kuvasarjat poikkesivat merkittävästi toisistaan kuvanlaadun ja pisaroiden ulkomuodon osalta, johtuen eroista kuvaus- ja hitsausprosesseissa. Analyysijärjestelmä kehitettiin toimimaan pienellä osajoukolla kuvasarjoja, joissa oli tietynlainen kuvaus- ja hitsausprosessi ja joiden kuvanlaatu ja pisaroiden ulkomuoto olivat samankaltaisia, mutta järjestelmää testattiin myös osajoukon ulkopuolisilla kuvasarjoilla. Testitulokset osoittivat, että lentosuunnanmääritystarkkuus oli kohtuullisen suuri osajoukonsisällä ja pieni muissa kuvasarjoissa. Valokaaren taajuuden säännöllisyyden määritys oli tarkka useammassa kuvasarjassa.