Pub Date : 2018-10-01DOI: 10.1109/SIBGRAPI.2018.00030
M. Bento, R. Souza, R. Frayne
Magnetic resonance (MR) as well as other imaging modalities have been used in a large number of clinical and research studies for the analysis and quantification of important structures and the detection of abnormalities. In this context, machine learning is playing an increasingly important role in the development of automated tools for aiding in image quantification, patient diagnosis and follow-up. Normally, these techniques require large, heterogeneous datasets to provide accurate and generalizable results. Large, multi-center studies, for example, can provide such data. Images acquired at different centers, however, can present varying characteristics due to differences in acquisition parameters, site procedures and scanners configuration. While variability in the dataset is required to develop robust, generalizable studies (i.e., independent of the acquisition parameters or center), like all studies there is also a need to ensure overall data quality by prospectively identifying and removing poor-quality data samples that should not be included, e.g., outliers. We wish to keep image samples that are representative of the underlying population (so called inliers), yet removing those samples that are not. We propose a framework to analyze data variability and identify samples that should be removed in order to have more representative, reliable and robust datasets. Our example case study is based on a public dataset containing T1-weighted volumetric head images data acquired at six different centers, using three different scanner vendors and at two commonly used magnetic fields strengths. We propose an algorithm for assessing data robustness and finding the optimal data for study occlusion (i.e., the data size that presents with lowest variability while maintaining generalizability (i.e., using samples from all sites)).
{"title":"Multicenter Imaging Studies: Automated Approach to Evaluating Data Variability and the Role of Outliers","authors":"M. Bento, R. Souza, R. Frayne","doi":"10.1109/SIBGRAPI.2018.00030","DOIUrl":"https://doi.org/10.1109/SIBGRAPI.2018.00030","url":null,"abstract":"Magnetic resonance (MR) as well as other imaging modalities have been used in a large number of clinical and research studies for the analysis and quantification of important structures and the detection of abnormalities. In this context, machine learning is playing an increasingly important role in the development of automated tools for aiding in image quantification, patient diagnosis and follow-up. Normally, these techniques require large, heterogeneous datasets to provide accurate and generalizable results. Large, multi-center studies, for example, can provide such data. Images acquired at different centers, however, can present varying characteristics due to differences in acquisition parameters, site procedures and scanners configuration. While variability in the dataset is required to develop robust, generalizable studies (i.e., independent of the acquisition parameters or center), like all studies there is also a need to ensure overall data quality by prospectively identifying and removing poor-quality data samples that should not be included, e.g., outliers. We wish to keep image samples that are representative of the underlying population (so called inliers), yet removing those samples that are not. We propose a framework to analyze data variability and identify samples that should be removed in order to have more representative, reliable and robust datasets. Our example case study is based on a public dataset containing T1-weighted volumetric head images data acquired at six different centers, using three different scanner vendors and at two commonly used magnetic fields strengths. We propose an algorithm for assessing data robustness and finding the optimal data for study occlusion (i.e., the data size that presents with lowest variability while maintaining generalizability (i.e., using samples from all sites)).","PeriodicalId":208985,"journal":{"name":"2018 31st SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114887915","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/SIBGRAPI.2018.00037
A. Moreira, Paulo Ivson, Waldemar Celes Filho
The recent advances in cloud services enable an increasing number of applications to offload their intensive tasks to remote computers. Cloud rendering comprises a set of services capable of rendering a 3D scene on a remote workstation. Notable progress in this field has been made by cloud gaming services. However, a gap remains between existing cloud rendering systems and other graphics-intensive applications, such as visualization of Computer-Aided Design (CAD) models. Existing cloud gaming services are not suitable to efficiently render these particular 3D scenes. CAD models contain many more objects than a regular game scene, requiring specific assumptions and optimizations to deliver an interactive user experience. In this work, we discuss and propose a novel hybrid cloud rendering system for massive 3D CAD models of industrial plants. The obtained results show that our technique can achieve high frame rates with satisfactory image quality even in a constrained environment, such as a high latency network or obsolete computer hardware.
{"title":"Hybrid Cloud Rendering System for Massive CAD Models","authors":"A. Moreira, Paulo Ivson, Waldemar Celes Filho","doi":"10.1109/SIBGRAPI.2018.00037","DOIUrl":"https://doi.org/10.1109/SIBGRAPI.2018.00037","url":null,"abstract":"The recent advances in cloud services enable an increasing number of applications to offload their intensive tasks to remote computers. Cloud rendering comprises a set of services capable of rendering a 3D scene on a remote workstation. Notable progress in this field has been made by cloud gaming services. However, a gap remains between existing cloud rendering systems and other graphics-intensive applications, such as visualization of Computer-Aided Design (CAD) models. Existing cloud gaming services are not suitable to efficiently render these particular 3D scenes. CAD models contain many more objects than a regular game scene, requiring specific assumptions and optimizations to deliver an interactive user experience. In this work, we discuss and propose a novel hybrid cloud rendering system for massive 3D CAD models of industrial plants. The obtained results show that our technique can achieve high frame rates with satisfactory image quality even in a constrained environment, such as a high latency network or obsolete computer hardware.","PeriodicalId":208985,"journal":{"name":"2018 31st SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI)","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126415386","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/SIBGRAPI.2018.00012
Matheus Macedo Leonardo, Tiago J. Carvalho, Edmar R. S. Rezende, R. Zucchi, F. Faria
Fruit flies has a big biological and economic importance for the farming of different tropical and subtropical countries in the World. Specifically in Brazil, third largest fruit producer in the world, the direct and indirect losses caused by fruit flies can exceed USD 120 million/year. These losses are related to production, the cost of pest control and export markets. One of the most economically important fruit flies in the America belong to the genus Anastrepha, which has approximately 300 known species, of which 120 are recorded in Brazil. However, less than 10 species are economically important and are considered pests of quarantine significance by regulatory agencies. The extreme similarity among the species of the genus Anastrepha makes its manual taxonomic classification a nontrivial task, causing onerous and very subjective results. In this work, we propose an approach based on deep learning to assist the scarce specialists, reducing the time of analysis, subjectivity of the classifications and consequently, the economic losses related to these agricultural pests. In our experiments, five deep features and nine machine learning techniques have been studied for the target task. Furthermore, the proposed approach have achieved similar effectiveness results to state-of-art approaches.
{"title":"Deep Feature-Based Classifiers for Fruit Fly Identification (Diptera: Tephritidae)","authors":"Matheus Macedo Leonardo, Tiago J. Carvalho, Edmar R. S. Rezende, R. Zucchi, F. Faria","doi":"10.1109/SIBGRAPI.2018.00012","DOIUrl":"https://doi.org/10.1109/SIBGRAPI.2018.00012","url":null,"abstract":"Fruit flies has a big biological and economic importance for the farming of different tropical and subtropical countries in the World. Specifically in Brazil, third largest fruit producer in the world, the direct and indirect losses caused by fruit flies can exceed USD 120 million/year. These losses are related to production, the cost of pest control and export markets. One of the most economically important fruit flies in the America belong to the genus Anastrepha, which has approximately 300 known species, of which 120 are recorded in Brazil. However, less than 10 species are economically important and are considered pests of quarantine significance by regulatory agencies. The extreme similarity among the species of the genus Anastrepha makes its manual taxonomic classification a nontrivial task, causing onerous and very subjective results. In this work, we propose an approach based on deep learning to assist the scarce specialists, reducing the time of analysis, subjectivity of the classifications and consequently, the economic losses related to these agricultural pests. In our experiments, five deep features and nine machine learning techniques have been studied for the target task. Furthermore, the proposed approach have achieved similar effectiveness results to state-of-art approaches.","PeriodicalId":208985,"journal":{"name":"2018 31st SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134416818","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/SIBGRAPI.2018.00046
Natal Henrique Cordeiro, E. C. Pedrino
The production of sensory substitution equipment for the visually impaired (VIP) is growing. The aim of this project is to understand the VIP context and predict the risks of collision for the VIP, following an analysis of the position, distance, size and motion of the objects present in their environment. This understanding is refined by data fusion steps applied to the Situation Awareness model to predict possible impacts in the near future. With this goal, a new architecture was designed, composed of systems that detect free passages, static objects, dynamic objects and the paths of these dynamic objects. The detected data was mapped into a 3D plane verifying positions and sizes. For the fusion, a method was developed that compared four more general classifiers in order to verify which presented greater reliability in the given context. These classifiers allowed inferences to be made when analyzing the risks of collision in different directions. The architecture designed for risk prediction is the main contribution of this project.
{"title":"An Architecture for Collision Risk Prediction for Visually Impaired People","authors":"Natal Henrique Cordeiro, E. C. Pedrino","doi":"10.1109/SIBGRAPI.2018.00046","DOIUrl":"https://doi.org/10.1109/SIBGRAPI.2018.00046","url":null,"abstract":"The production of sensory substitution equipment for the visually impaired (VIP) is growing. The aim of this project is to understand the VIP context and predict the risks of collision for the VIP, following an analysis of the position, distance, size and motion of the objects present in their environment. This understanding is refined by data fusion steps applied to the Situation Awareness model to predict possible impacts in the near future. With this goal, a new architecture was designed, composed of systems that detect free passages, static objects, dynamic objects and the paths of these dynamic objects. The detected data was mapped into a 3D plane verifying positions and sizes. For the fusion, a method was developed that compared four more general classifiers in order to verify which presented greater reliability in the given context. These classifiers allowed inferences to be made when analyzing the risks of collision in different directions. The architecture designed for risk prediction is the main contribution of this project.","PeriodicalId":208985,"journal":{"name":"2018 31st SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI)","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133336310","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/SIBGRAPI.2018.00053
F. Dias, R. Minghim
The complexity and size of data have created challenges to data analysis. Although point placement strategies have gained popularity in the last decade to yield a global view of multidimensional datasets, few attempts have been made to improve visual scalability and offer multilevel exploration in the context of multidimensional projections and point placement strategies. Such approaches can be helpful in improving the analysis capability both by organizing visual spaces and allowing meaningful partitions of larger datasets. In this paper, we extend the Hierarchy Point Placement (HiPP), a strategy for multi-level point placement, in order to enhance its analytical capabilities and flexibility to handle current trends in visual data science. We have provided several combinations of clustering methods and projection approaches to represent and visualize datasets; added a choice to invert the original processing order from cluster-projection to projection-cluster; proposed a better way to initialize the partitions, and added ways to summarize image, audio, text and general data groups. The tool's code is made available online. In this article, we present the new tool (named xHiPP) and provide validation through case studies with simpler and more complex datasets (text and audio) to illustrate that the capabilities afforded by the extensions have managed to provide analysts with the ability to quickly gain insight and adjust the processing pipeline to their needs.
{"title":"xHiPP: eXtended Hierarchical Point Placement Strategy","authors":"F. Dias, R. Minghim","doi":"10.1109/SIBGRAPI.2018.00053","DOIUrl":"https://doi.org/10.1109/SIBGRAPI.2018.00053","url":null,"abstract":"The complexity and size of data have created challenges to data analysis. Although point placement strategies have gained popularity in the last decade to yield a global view of multidimensional datasets, few attempts have been made to improve visual scalability and offer multilevel exploration in the context of multidimensional projections and point placement strategies. Such approaches can be helpful in improving the analysis capability both by organizing visual spaces and allowing meaningful partitions of larger datasets. In this paper, we extend the Hierarchy Point Placement (HiPP), a strategy for multi-level point placement, in order to enhance its analytical capabilities and flexibility to handle current trends in visual data science. We have provided several combinations of clustering methods and projection approaches to represent and visualize datasets; added a choice to invert the original processing order from cluster-projection to projection-cluster; proposed a better way to initialize the partitions, and added ways to summarize image, audio, text and general data groups. The tool's code is made available online. In this article, we present the new tool (named xHiPP) and provide validation through case studies with simpler and more complex datasets (text and audio) to illustrate that the capabilities afforded by the extensions have managed to provide analysts with the ability to quickly gain insight and adjust the processing pipeline to their needs.","PeriodicalId":208985,"journal":{"name":"2018 31st SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131365914","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/SIBGRAPI.2018.00018
T. M. Paixão, Rodrigo Berriel, M. C. Boeres, C. Badue, A. D. Souza, Thiago Oliveira-Santos
The use of paper-shredder machines (mechanical shredding) to destroy documents can be illicitly motivated when the purpose is hiding evidence of fraud and other sorts of crimes. Therefore, reconstructing such documents is of great value for forensic investigation, but it is admittedly a stressful and time-consuming task for humans. To address this challenge, several computational techniques have been proposed in literature, particularly for documents with text-based content. In this context, a critical challenge for automated reconstruction is to measure properly the fitting (compatibility) between paper shreds (strips), which has been observed to be the main limitation of literature on this topic. The main contribution of this paper is a deep learning-based compatibility score to be applied in the reconstruction of strip-shredded text documents. Since there is no abundance of real-shredded data, we propose a training scheme based on digital simulated-shredding of documents from a well-known OCR database. The proposed score was coupled to a black-box optimization tool, and the resulting system achieved an average accuracy of 94.58% in the reconstruction of mechanically-shredded documents.
{"title":"A Deep Learning-Based Compatibility Score for Reconstruction of Strip-Shredded Text Documents","authors":"T. M. Paixão, Rodrigo Berriel, M. C. Boeres, C. Badue, A. D. Souza, Thiago Oliveira-Santos","doi":"10.1109/SIBGRAPI.2018.00018","DOIUrl":"https://doi.org/10.1109/SIBGRAPI.2018.00018","url":null,"abstract":"The use of paper-shredder machines (mechanical shredding) to destroy documents can be illicitly motivated when the purpose is hiding evidence of fraud and other sorts of crimes. Therefore, reconstructing such documents is of great value for forensic investigation, but it is admittedly a stressful and time-consuming task for humans. To address this challenge, several computational techniques have been proposed in literature, particularly for documents with text-based content. In this context, a critical challenge for automated reconstruction is to measure properly the fitting (compatibility) between paper shreds (strips), which has been observed to be the main limitation of literature on this topic. The main contribution of this paper is a deep learning-based compatibility score to be applied in the reconstruction of strip-shredded text documents. Since there is no abundance of real-shredded data, we propose a training scheme based on digital simulated-shredding of documents from a well-known OCR database. The proposed score was coupled to a black-box optimization tool, and the resulting system achieved an average accuracy of 94.58% in the reconstruction of mechanically-shredded documents.","PeriodicalId":208985,"journal":{"name":"2018 31st SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI)","volume":"60 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131488984","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/SIBGRAPI.2018.00028
L. Souza, A. Ebigbo, A. Probst, H. Messmann, J. Papa, R. Mendel, C. Palm
In this work, we propose the use of single channel Color Co-occurrence Matrices for texture description of Barrett's Esophagus (BE) and adenocarcinoma images. Further classification using supervised learning techniques, such as Optimum-Path Forest (OPF), Support Vector Machines with Radial Basis Function (SVM-RBF) and Bayesian classifier supports the context of automatic BE and adenocarcinoma diagnosis. We validated three approaches of classification based on patches, patients and images in two datasets (MICCAI 2015 and Augsburg) using the color-and-texture descriptors and the machine learning techniques. Concerning MICCAI 2015 dataset, the best results were obtained using the blue channel for the descriptors and the supervised OPF for classification purposes in the patch-based approach, with sensitivity nearly to 73% for positive adenocarcinoma identification and specificity close to 77% for BE (non-cancerous) patch classification. Regarding the Augsburg dataset, the most accurate results were also obtained using both OPF classifier and blue channel descriptor for the feature extraction, with sensitivity close to 67% and specificity around to 76%. Our work highlights new advances in the related research area and provides a promising technique that combines color and texture information, allied to three different approaches of dataset pre-processing aiming to configure robust scenarios for the classification step.
{"title":"Barrett's Esophagus Identification Using Color Co-Occurrence Matrices","authors":"L. Souza, A. Ebigbo, A. Probst, H. Messmann, J. Papa, R. Mendel, C. Palm","doi":"10.1109/SIBGRAPI.2018.00028","DOIUrl":"https://doi.org/10.1109/SIBGRAPI.2018.00028","url":null,"abstract":"In this work, we propose the use of single channel Color Co-occurrence Matrices for texture description of Barrett's Esophagus (BE) and adenocarcinoma images. Further classification using supervised learning techniques, such as Optimum-Path Forest (OPF), Support Vector Machines with Radial Basis Function (SVM-RBF) and Bayesian classifier supports the context of automatic BE and adenocarcinoma diagnosis. We validated three approaches of classification based on patches, patients and images in two datasets (MICCAI 2015 and Augsburg) using the color-and-texture descriptors and the machine learning techniques. Concerning MICCAI 2015 dataset, the best results were obtained using the blue channel for the descriptors and the supervised OPF for classification purposes in the patch-based approach, with sensitivity nearly to 73% for positive adenocarcinoma identification and specificity close to 77% for BE (non-cancerous) patch classification. Regarding the Augsburg dataset, the most accurate results were also obtained using both OPF classifier and blue channel descriptor for the feature extraction, with sensitivity close to 67% and specificity around to 76%. Our work highlights new advances in the related research area and provides a promising technique that combines color and texture information, allied to three different approaches of dataset pre-processing aiming to configure robust scenarios for the classification step.","PeriodicalId":208985,"journal":{"name":"2018 31st SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI)","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121739294","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/SIBGRAPI.2018.00023
K. Cunha, Lucas Maggi, V. Teichrieb, J. P. Lima, J. Quintino, F. Q. Silva, André L. M. Santos, Helder Pinho
In this work we address the problem of landmark recognition. We extend PlaNet, a model based on deep neural networks that approaches the problem of landmark recognition as a classification problem and performs the recognition of places around the world. We propose an extension of the PlaNet technique in which we use a voting scheme to perform the classification, dividing the image into previously defined regions and inferring the landmark based on these regions. The prediction of the model depends not only on the information of the features learned by the deep convolutional neural network architecture during training, but also uses local information from each region in the image for which the classification is made. To validate our proposal, we performed the training of the original PlaNet model and our variation using a database built with images from Flickr, and evaluated the models in the Paris and Oxford Buildings datasets. It was possible to notice that the addition of image division and voting structure improves the accuracy result of the model by 5-11 percentage points on average, reducing the level of ambiguity found during the inference of the model.
{"title":"Patch PlaNet: Landmark Recognition with Patch Classification Using Convolutional Neural Networks","authors":"K. Cunha, Lucas Maggi, V. Teichrieb, J. P. Lima, J. Quintino, F. Q. Silva, André L. M. Santos, Helder Pinho","doi":"10.1109/SIBGRAPI.2018.00023","DOIUrl":"https://doi.org/10.1109/SIBGRAPI.2018.00023","url":null,"abstract":"In this work we address the problem of landmark recognition. We extend PlaNet, a model based on deep neural networks that approaches the problem of landmark recognition as a classification problem and performs the recognition of places around the world. We propose an extension of the PlaNet technique in which we use a voting scheme to perform the classification, dividing the image into previously defined regions and inferring the landmark based on these regions. The prediction of the model depends not only on the information of the features learned by the deep convolutional neural network architecture during training, but also uses local information from each region in the image for which the classification is made. To validate our proposal, we performed the training of the original PlaNet model and our variation using a database built with images from Flickr, and evaluated the models in the Paris and Oxford Buildings datasets. It was possible to notice that the addition of image division and voting structure improves the accuracy result of the model by 5-11 percentage points on average, reducing the level of ambiguity found during the inference of the model.","PeriodicalId":208985,"journal":{"name":"2018 31st SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI)","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122366858","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/SIBGRAPI.2018.00008
L. Figueiredo, Paulo Ivson, Waldemar Celes Filho
3D CAD models are widely used to improve management of large-scale engineering projects. Examples include Building Information Modeling (BIM) and Oil & Gas industrial plants. Maintaining these facilities is a critical task that often involves anti-corrosive painting of equipment and metallic structures. Existing CAD software estimates the painting area including hidden surfaces that are not actually painted in the field. To improve these computations, we propose an approach based on Adaptively-Sampled Distance Fields (ADFs) exploiting the relationship between object areas and Constructive Solid Geometry (CSG) operations. Tests with synthetic models demonstrate that our technique achieves an accuracy of 99%. In real-world 3D CAD models, we were able to reduce the estimated area by 38% when compared to the naïve calculations. These result in significant cost savings in material provision and workforce required for maintaining these facilities.
{"title":"Hidden Surface Removal for Accurate Painting-Area Calculation on CAD Models","authors":"L. Figueiredo, Paulo Ivson, Waldemar Celes Filho","doi":"10.1109/SIBGRAPI.2018.00008","DOIUrl":"https://doi.org/10.1109/SIBGRAPI.2018.00008","url":null,"abstract":"3D CAD models are widely used to improve management of large-scale engineering projects. Examples include Building Information Modeling (BIM) and Oil & Gas industrial plants. Maintaining these facilities is a critical task that often involves anti-corrosive painting of equipment and metallic structures. Existing CAD software estimates the painting area including hidden surfaces that are not actually painted in the field. To improve these computations, we propose an approach based on Adaptively-Sampled Distance Fields (ADFs) exploiting the relationship between object areas and Constructive Solid Geometry (CSG) operations. Tests with synthetic models demonstrate that our technique achieves an accuracy of 99%. In real-world 3D CAD models, we were able to reduce the estimated area by 38% when compared to the naïve calculations. These result in significant cost savings in material provision and workforce required for maintaining these facilities.","PeriodicalId":208985,"journal":{"name":"2018 31st SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131600618","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/SIBGRAPI.2018.00021
G. Gonçalves, M. A. Diniz, Rayson Laroca, D. Menotti, W. R. Schwartz
With the increasing number of cameras available in the cities, video traffic analysis can provide useful insights for the transportation segment. One of such analysis is the Automatic License Plate Recognition (ALPR). Previous approaches divided this task into several cascaded subtasks, i.e., vehicle location, license plate detection, character segmentation and optical character recognition. However, since each task has its own accuracy, the error propagation between each subtask is detrimental to the final accuracy. Therefore, focusing on the reduction of error propagation, we propose a technique that is able to perform ALPR using only two deep networks, the first performs license plate detection (LPD) and the second performs license plate recognition (LPR). The latter does not execute explicit character segmentation, which reduces significantly the error propagation. As these deep networks need a large number of samples to converge, we develop new data augmentation techniques that allow them to reach their full potential as well as a new dataset to train and evaluate ALPR approaches. According to experimental results, our approach is able to achieve state-of-the-art results in the SSIG-SegPlate dataset, reaching improvements up to 1.4 percentage point when compared to the best baseline. Furthermore, the approach is also able to perform in real time even in scenarios where many plates are present at the same frame, reaching significantly higher frame rates when compared with previously proposed approaches.
{"title":"Real-Time Automatic License Plate Recognition through Deep Multi-Task Networks","authors":"G. Gonçalves, M. A. Diniz, Rayson Laroca, D. Menotti, W. R. Schwartz","doi":"10.1109/SIBGRAPI.2018.00021","DOIUrl":"https://doi.org/10.1109/SIBGRAPI.2018.00021","url":null,"abstract":"With the increasing number of cameras available in the cities, video traffic analysis can provide useful insights for the transportation segment. One of such analysis is the Automatic License Plate Recognition (ALPR). Previous approaches divided this task into several cascaded subtasks, i.e., vehicle location, license plate detection, character segmentation and optical character recognition. However, since each task has its own accuracy, the error propagation between each subtask is detrimental to the final accuracy. Therefore, focusing on the reduction of error propagation, we propose a technique that is able to perform ALPR using only two deep networks, the first performs license plate detection (LPD) and the second performs license plate recognition (LPR). The latter does not execute explicit character segmentation, which reduces significantly the error propagation. As these deep networks need a large number of samples to converge, we develop new data augmentation techniques that allow them to reach their full potential as well as a new dataset to train and evaluate ALPR approaches. According to experimental results, our approach is able to achieve state-of-the-art results in the SSIG-SegPlate dataset, reaching improvements up to 1.4 percentage point when compared to the best baseline. Furthermore, the approach is also able to perform in real time even in scenarios where many plates are present at the same frame, reaching significantly higher frame rates when compared with previously proposed approaches.","PeriodicalId":208985,"journal":{"name":"2018 31st SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125955270","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：