The frequent growth of visual data, either by countless available monitoring video cameras or the... more The frequent growth of visual data, either by countless available monitoring video cameras or the popularization of mobile devices that allow each person to create, edit, and share their own images and videos have contributed enormously to the so-called "big-data revolution". This shear amount of visual data gives rise to a Pandora box of new visual classification problems never imagined before. Image and video classification tasks have been inserted in different and complex applications and the use of machine learning-based solutions has become the most popular approach to several applications. Notwithstanding, there is no silver bullet that solves all the problems, i.e., it is not possible to characterize all images of different domains with the same description method nor is it possible to use the same learning method to achieve good results in any kind of application. In this thesis, we aim at proposing a framework for classifier selection and fusion. Our method seeks to combine image characterization and learning methods by means of a meta-learning approach responsible for assessing which methods contribute more towards the solution of a given problem. The framework uses three different strategies of classifier selection which pinpoints the less correlated, yet effective, classifiers through a series of diversity measure analysis. The experiments show that the proposed approaches yield comparable results to well-known algorithms from the literature on many different applications but using less learning and description methods as well as not incurring in the curse of dimensionality and normalization problems common to some fusion techniques. Furthermore, our approach is able to achieve effective classification results using very reduced training sets. To my family and friends. Para minha família e amigos. Eu gostaria de agradecer ao meu pai pelo suporte, pela confiança depositada em todos os momentos, por ser essa pessoa tão importante em minha vida e que sou grato todos os dias (um grande amigo); aos meus irmãos Sérgio e Vitor pelo apoio e amizade verdadeira que temos; à minha princesa e quase esposa Ariadne pelo carinho, paciência e cumplicidade em todos os momentos; aos meus colegas de Laboratório (Recod) do Instituto de Computação da Unicamp pelo ambiente descontraído e pelas colaborações que tivemos durante o período; a todos os amigos por existirem, pois sem eles eu nada seria. Agradeço ao orientador Prof. Dr. Ricardo da Silva Torres pela paciência, compreensão, ensinamentos e principalmente por se lembrar que pessoas são diferentes e por isso, necessitam de tratamentos diferentes. Esta capacidade de gerenciar pessoas é fundamental para seu sucesso e não tenho dúvidas que só aumentará no futuro. Deixo aqui o meu sincero MUITO OBRIGADO! Agradeço ao co-orientador Prof. Dr. Anderson Rocha pelos ensinamentos e por todas as discussões "acaloradas" que tivemos durante o período. Tenho certeza que cada reunião de pesquisa me fez crescer ainda mais como pessoa, professor e pesquisador; à Prof. Dra. Claudia Medeiros pelo apoio, sabedoria e pela oportunidade de fazer parte de seu grupo de pesquisas (LIS); a todos colaboradores de diferentes universidades do país e fora dele, pela parceria nos trabalhos realizados e nos que ainda virão. Agradeço aos funcionários do Instituto de Computação, principalmente Daniel, Flavio, Wilson e Dona Rosa que foram sempre muito dedicados aos alunos de pós-graduação. Agradeço a Fundação de Amparo à Pesquisa do Estado de São Paulo -FAPESP (processo 2010/14910-0 ), Conselho Nacional de Desenvolvimento Científico e Tecnológico -CNPq e Coordenação de Aperfeiçoamento de Pessoal de Nivel Superior -CAPES (proceso 1260-12-0) pelo apoio financeiro. I would like to thank Prof. Dr. Sudeep Sarkar for all the wonderful things he did for me while I was at University of South Florida (USF) and by treatment that I have received there. I have no words to describe how I am grateful to him for accepting me in his research group. The experience I gained will be extremely important to my personal and professional life. xv Epigraph "The more I see of the World, the more I find out the extent of my ignorance." Just paraphrasing Socrates... Epígrafe "Quanto mais conheço o Mundo, mais vejo o tamanho da minha ignorância." Apenas parafraseando Socrates...
This paper addresses the problem of remote sensing image multi-scale classification by:(i) showin... more This paper addresses the problem of remote sensing image multi-scale classification by:(i) showing that using multiple scales does improve classification results, but not all scales have the same importance;(ii) showing that image descriptors do not offer the same contribution at all scales, as commonly thought, and some of them are very correlated;(iii) introducing a simple approach to automatically select segmentation scales, descriptors, and classifiers based on correlation and accuracy analysis.
2022 35th SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI)
Tropical forests represent the home of many species on the planet for flora and fauna, retaining ... more Tropical forests represent the home of many species on the planet for flora and fauna, retaining billions of tons of carbon footprint, promoting clouds and rain formation, implying a crucial role in the global ecosystem, besides representing the home to countless indigenous peoples. Unfortunately, millions of hectares of tropical forests are lost every year due to deforestation or degradation. To mitigate that fact, monitoring and deforestation detection programs are in use, in addition to public policies for the prevention and punishment of criminals. These monitoring/detection programs generally use remote sensing images, image processing techniques, machine learning methods, and expert photointerpretation to analyze, identify and quantify possible changes in forest cover. Several projects have proposed different computational approaches, tools, and models to efficiently identify recent deforestation areas, improving deforestation monitoring programs in tropical forests. In this sense, this paper proposes the use of pattern classifiers based on neuroevolution technique (NEAT) in tropical forest deforestation detection tasks. Furthermore, a novel framework called e-NEAT has been created and achieved classification results above 90% for balanced accuracy measure in the target application using an extremely reduced and limited training set for learning the classification models. These results represent a relative gain of 6.2% over the best baseline ensemble method compared in this paper .
Research on stomata, i.e., morphological structures of plants, has increased in popularity in the... more Research on stomata, i.e., morphological structures of plants, has increased in popularity in the last years. These structures (pores) are in charge of the interaction between the internal plant system and the environment, working on different processes such as photosynthesis and transpiration stream. Besides, a better understanding of the pore mechanism plays a significant role when exploring the evolution process, as well as the behavior of plants. Although the study of stomata in dicots species of plants has advanced considerably in the past years, there is little information about stomata of cereal grasses. Also, automated detection of these structures have been considered in the literature, but some gaps are still uncovered. This fact is motivated by high morphological variation of stomata and the presence of noise from the image acquisition step. In this work, we propose a new methodology for automatic stomata classification and a new detection system in microscope images for ...
Uploads
Papers by fabio faria