A central task in computer vision is detecting object classes such as cars and horses in complex scenes. Training an object class detector typically requires a large set of images labeled with tight bounding boxes around every object instance. Obtaining such data requires human annotation, which is very expensive and time consuming. Alternatively, researchers have tried to train models in a weakly supervised setting (i.e., given only image-level labels), which is much cheaper but leads to weaker detectors. In this thesis, we propose new and efficient human annotation schemes for training object class detectors that bypass the need for drawing bounding boxes and reduce the annotation cost while still obtaining high quality object detectors. First, we propose to train object class detectors from eye tracking data. Instead of drawing tight bounding boxes, the annotators only need to look at the image and find the target object. We track the eye movements of annotators while they perform this visual search task and we propose a technique for deriving object bounding boxes from these eye fixations. To validate our idea, we augment an existing object detection dataset with eye tracking data. Second, we propose a scheme for training object class detectors, which only requires annotators to verify bounding-boxes produced automatically by the learning algorithm. Our scheme introduces human verification as a new step into a standard weakly supervised framework which typically iterates between re-training object detectors and re-localizing objects in the training images. We use the verification signal to improve both re-training and re-localization. Third, we propose another scheme where annotators are asked to click on the center of an imaginary bounding box, which tightly encloses the object. We then incorporate these clicks into a weakly supervised object localization technique, to jointly localize object bounding boxes over all training images. Both our center-clicking and human verification schemes deliver detectors performing almost as well as those trained in a fully supervised setting. Finally, we propose extreme clicking. We ask the annotator to click on four physical points on the object: the top, bottom, left- and right-most points. This task is more natural than the traditional way of drawing boxes and these points are easy to find. Our experiments show that annotating objects with extreme clicking is 5 X faster than the traditional way of drawing boxes and it leads to boxes of the same quality as the original ground-truth drawn the traditional way. Moreover, we use the resulting extreme points to obtain more accurate segmentations than those derived from bounding boxes.

A Disfunção Temporomandibular (DTM) se refere a várias doenças que envolvem os músculos da mastigação e/ou a articulação temporomandibular (ATM). Os distúrbios de desarranjo do disco são considerados uma subdivisão dos distúrbios intra-articulares da ATM e são aqueles onde há relacionamento anormal do disco articular com a cabeça da mandíbula, fossa articular e eminência articular. O procedimento para diagnosticar a DTM geralmente consiste em exame clínico, mas quando informações adicionais são necessárias para confirmar esse resultado, exames de imagem são solicitados. Imagem por ressonância magnética (IRM) é considerada o exame padrão-ouro de referência de diagnóstico não invasivo de deslocamento de disco. Na literatura revisada não há justificativa de que o corte central da IRM deva ser o utilizado para classificar o posicionamento do disco e talvez o uso de vários cortes da IRM pudesse fornecer um dado importante para uma correta interpretação da posição do disco. O presente estudo avaliou a concordância entre três avaliadores para ambos os critérios de classificação (corte central versus vários cortes); verificou a concordância entre os dois critérios, além de correlacionar os resultados obtidos por ambos os critérios com o sinal clínico de estalo em pacientes diagnosticados com DTM. Para medir os níveis de concordância das proporções de imagens deslocadas entre os três avaliadores foi utilizado o coeficiente de correlação intraclasse (ICC). Os níveis de concordância entre os métodos e entre as avaliadoras foram medidos através do método Kappa. O teste utilizado para verificar associações entre as variáveis foi o teste exato de Fisher bicaudal e o nível de significância adotado foi de 0,05. Do total de pacientes examinados 70% (42/60) eram do sexo feminino e 30% (18/60) do sexo masculino. A média de idade foi de 37,2 ± 12,5, Mensurando a concordância entre os avaliadores para os critérios diagnósticos do posicionamento do disco articular da ATM o ICC obtido foi excelente (0,839) para o critério que considerou a proporção de discos deslocados. Para o critério de pelo menos um disco deslocado um Kappa substancial de 0,795 foi alcançado e para o critério do corte central um Kappa de concordância quase perfeita de 0,806 foi encontrado. Um Kappa alto (0,815) também foi atingido na concordância da classificação do posicionamento do disco na IRM em abertura. Após os critérios de avaliação para a posição do disco serem submetidos aos testes estatísticos para a averiguação entre a concordância deles, um Kappa de concordância substancial de 0,697 foi constatado. Notou-se nesse estudo que existe uma associação entre o estalo e a classificação com o disco deslocado pelo método da corte central (p=0,01) e entre o diagnóstico de deslocamento do disco articular obtido através da IMR em abertura da ATM e o estalo (p=0,011). Não foi encontrada associação entre o estalo articular e a posição do disco articular pelo critério de vários cortes. A diferença entre os resultados obtidos pelos dois critérios de avaliação justifica a proposta de que os pesquisadores deveriam descrever qual o critério utilizado em seu estudo para propiciar uma melhor comparação entre as literaturas. / The Temporomandibular Dysfunction (TMD) refers to several diseases related to the muscles involved in chewing and/or the temporomandibular joint (TMJ). The articular disc derangement disorders are considered a subdivision of TMJ intra-articular disorders and are the ones on which there is an abnormal relationship between the joint disc, the condyle (head of the jaw), the articular fossae and the articular eminence. The procedure to diagnose temporomandibular dysfunction consists of five clinical examinations, but when additional information becomes necessary to confirm the results, image techniques are required. Magnetic Resonance Imaging (MRI) is considered the golden standard technique and is the reference for non-invasive diagnosis of disc displacement. In the literature review it is not conclusive that the central slice of the MRI must be used to classify the disc position, and it is possible that the use of several slices of MRI could provide important data for a more appropriate interpretation of the disc position. This study compared the agreement of the findings between three analyzers for both classification criteria (central slice and several slices); verified the agreement between the two criteria, as well as correlated the obtained results for both criteria with the clinical sign of clicking in patients diagnosed with TMD. In order to measure the agreement of the displacement image proportions between the three analyzers, the interclass correlation coefficient (ICC) was used. The degrees of agreement between the analyzers were measured with the Kappa method. The test used to verify the association between the variables was the two-tailed Fischer exact test and the adopted level of significance was 0.05. From the total patients examined, 70% (42/60) were females and 30% (18/60) were males. The average age was 37,2 ± 12,5. The measurements for the agreement between the analyzers for the diagnosis criteria of the joint disc position of the TMJ resulted in an excellent ICC (0,839) for the criteria that considered the disc displacement proportions. For the criteria of at least one displaced disc, a substantial Kappa of 0,795 was reached, and for the criteria of central slice an almost perfect Kappa agreement of 0,806 was reached. A high Kappa (0,815) was also obtained in the agreement of the disc placement classification in the opened mouth MRI. After the criteria of analysis for disc position were submitted to statistical tests for validation of the agreement between them, a Kappa of substantial agreement of 0,697 was registered. It was observed in this study that there exists an association between clicking and the classification of disc displacement by the central slice method (p=0,01) as well as between the joint disc displacement diagnosis obtained through the opened mouth MRI of the TMJ and the clicking (p=0,011). No association was found between the joint clicking and the joint disc position using the criteria of several slices. The difference between the results obtained by the two analysis criteria justifies the proposal that the researchers should describe which criteria was used in their study in order to enable a better comparison between the literatures.

Some species of hoofed animals produce clicking sound. This work analysed clicking parameters (dominant frequency, 25% quartile, 50% quartile, 75% quartile) of this species: Kuban Tur (Capra caucasica), Daghestan Tur (Capra cylindricornis) and Aoudad (Ammotragus lervia), which are clicking significantly. Clicks of each individual were recorded using solid state recorder Marantz PMD 620 with microphone. All clicks were analysed using program Avisoft-SAS Lab Pro Software, verze 5.0.01 (2010). Data were analysed using program STATISTICA, version 9 (ANONYMUS 2009). Individuals were compared using one-way ANOVA and discriminant analysis. Species and sexes were compared using discriminant analysis and nested ANOVA. Clicking parameters were correlated with age. A phylogenetic distribution of clicking was determined in Bovidae. Results suggest differences between individuals and species. Differences between sexes wasn't significant. Correlation with age was significant.

A Disfunção Temporomandibular (DTM) se refere a várias doenças que envolvem os músculos da mastigação e/ou a articulação temporomandibular (ATM). Os distúrbios de desarranjo do disco são considerados uma subdivisão dos distúrbios intra-articulares da ATM e são aqueles onde há relacionamento anormal do disco articular com a cabeça da mandíbula, fossa articular e eminência articular. O procedimento para diagnosticar a DTM geralmente consiste em exame clínico, mas quando informações adicionais são necessárias para confirmar esse resultado, exames de imagem são solicitados. Imagem por ressonância magnética (IRM) é considerada o exame padrão-ouro de referência de diagnóstico não invasivo de deslocamento de disco. Na literatura revisada não há justificativa de que o corte central da IRM deva ser o utilizado para classificar o posicionamento do disco e talvez o uso de vários cortes da IRM pudesse fornecer um dado importante para uma correta interpretação da posição do disco. O presente estudo avaliou a concordância entre três avaliadores para ambos os critérios de classificação (corte central versus vários cortes); verificou a concordância entre os dois critérios, além de correlacionar os resultados obtidos por ambos os critérios com o sinal clínico de estalo em pacientes diagnosticados com DTM. Para medir os níveis de concordância das proporções de imagens deslocadas entre os três avaliadores foi utilizado o coeficiente de correlação intraclasse (ICC). Os níveis de concordância entre os métodos e entre as avaliadoras foram medidos através do método Kappa. O teste utilizado para verificar associações entre as variáveis foi o teste exato de Fisher bicaudal e o nível de significância adotado foi de 0,05. Do total de pacientes examinados 70% (42/60) eram do sexo feminino e 30% (18/60) do sexo masculino. A média de idade foi de 37,2 ± 12,5, Mensurando a concordância entre os avaliadores para os critérios diagnósticos do posicionamento do disco articular da ATM o ICC obtido foi excelente (0,839) para o critério que considerou a proporção de discos deslocados. Para o critério de pelo menos um disco deslocado um Kappa substancial de 0,795 foi alcançado e para o critério do corte central um Kappa de concordância quase perfeita de 0,806 foi encontrado. Um Kappa alto (0,815) também foi atingido na concordância da classificação do posicionamento do disco na IRM em abertura. Após os critérios de avaliação para a posição do disco serem submetidos aos testes estatísticos para a averiguação entre a concordância deles, um Kappa de concordância substancial de 0,697 foi constatado. Notou-se nesse estudo que existe uma associação entre o estalo e a classificação com o disco deslocado pelo método da corte central (p=0,01) e entre o diagnóstico de deslocamento do disco articular obtido através da IMR em abertura da ATM e o estalo (p=0,011). Não foi encontrada associação entre o estalo articular e a posição do disco articular pelo critério de vários cortes. A diferença entre os resultados obtidos pelos dois critérios de avaliação justifica a proposta de que os pesquisadores deveriam descrever qual o critério utilizado em seu estudo para propiciar uma melhor comparação entre as literaturas. / The Temporomandibular Dysfunction (TMD) refers to several diseases related to the muscles involved in chewing and/or the temporomandibular joint (TMJ). The articular disc derangement disorders are considered a subdivision of TMJ intra-articular disorders and are the ones on which there is an abnormal relationship between the joint disc, the condyle (head of the jaw), the articular fossae and the articular eminence. The procedure to diagnose temporomandibular dysfunction consists of five clinical examinations, but when additional information becomes necessary to confirm the results, image techniques are required. Magnetic Resonance Imaging (MRI) is considered the golden standard technique and is the reference for non-invasive diagnosis of disc displacement. In the literature review it is not conclusive that the central slice of the MRI must be used to classify the disc position, and it is possible that the use of several slices of MRI could provide important data for a more appropriate interpretation of the disc position. This study compared the agreement of the findings between three analyzers for both classification criteria (central slice and several slices); verified the agreement between the two criteria, as well as correlated the obtained results for both criteria with the clinical sign of clicking in patients diagnosed with TMD. In order to measure the agreement of the displacement image proportions between the three analyzers, the interclass correlation coefficient (ICC) was used. The degrees of agreement between the analyzers were measured with the Kappa method. The test used to verify the association between the variables was the two-tailed Fischer exact test and the adopted level of significance was 0.05. From the total patients examined, 70% (42/60) were females and 30% (18/60) were males. The average age was 37,2 ± 12,5. The measurements for the agreement between the analyzers for the diagnosis criteria of the joint disc position of the TMJ resulted in an excellent ICC (0,839) for the criteria that considered the disc displacement proportions. For the criteria of at least one displaced disc, a substantial Kappa of 0,795 was reached, and for the criteria of central slice an almost perfect Kappa agreement of 0,806 was reached. A high Kappa (0,815) was also obtained in the agreement of the disc placement classification in the opened mouth MRI. After the criteria of analysis for disc position were submitted to statistical tests for validation of the agreement between them, a Kappa of substantial agreement of 0,697 was registered. It was observed in this study that there exists an association between clicking and the classification of disc displacement by the central slice method (p=0,01) as well as between the joint disc displacement diagnosis obtained through the opened mouth MRI of the TMJ and the clicking (p=0,011). No association was found between the joint clicking and the joint disc position using the criteria of several slices. The difference between the results obtained by the two analysis criteria justifies the proposal that the researchers should describe which criteria was used in their study in order to enable a better comparison between the literatures.

It is well known that some ungulates produce clicking sound by limbs, but often without more details. This study deals with the clicking sound in five cervid taxons, specifically in Caribou (Rangifer tarandus), Pere David´s Deer (Elaphurus davidianus), Western Red Deer (Cervus elaphus), Moose (Alces alces), Barbary stag (Cervus elaphus barbarus) in detail. Dominant frequency, 25% quartile, 50% quartile, 75% quartile were investigated for these species, with the particular attention to sex, age and limb position (forelimb, hind-limb) of studied individuals. Clicks sounds were recorded by solid state recorder Marantz PMD 620 with microphone and handset and analysed using program Avisoft-SAS Lab Pro Software, verze 5.0.01 (2010). Statistical analysis of obtained sound parameters were performed using program STATISTICA, version 12 (ANONYMUS 2012).Individuals within the species were compared using discriminant analysis and one-way ANOVA, species using discriminant analysis and nested ANOVA. Nested ANOVA was also used for testing of the sex and age influence on click parameters. Phylogenetic distribution of clicking was determined using the parsimony approach. Results suggest differences among within particular taxon, but differences among species depend on used methods (they are different using nested ANOVA, but they are not different except for Barbary stag in discriminant analysis). Click parameters were significantly associated with age, but not with sex of particular individuals. Ancestor of cervids seems to produce clicking, the evolution of this sound was associated later with its reduction in course of cervid evolution.