A carregar...
Projeto de investigação
IPSTERS - IPSentinel Terrestrial Enhanced Recognition System
Financiador
Autores
Publicações
Improving imbalanced land cover classification with k-means smote
Publication . Fonseca, Joao; Douzas, Georgios; Bacao, Fernando; Information Management Research Center (MagIC) - NOVA Information Management School; NOVA Information Management School (NOVA IMS); MDPI - Multidisciplinary Digital Publishing Institute
Land cover maps are a critical tool to support informed policy development, planning, and resource management decisions. With significant upsides, the automatic production of Land Use/Land Cover maps has been a topic of interest for the remote sensing community for several years, but it is still fraught with technical challenges. One such challenge is the imbalanced nature of most remotely sensed data. The asymmetric class distribution impacts negatively the performance of classifiers and adds a new source of error to the production of these maps. In this paper, we address the imbalanced learning problem, by using K-means and the Synthetic Minority Oversampling Technique (SMOTE) as an improved oversampling algorithm. K-means SMOTE improves the quality of newly created artificial data by addressing both the between-class imbalance, as traditional oversamplers do, but also the within-class imbalance, avoiding the generation of noisy data while effectively overcoming data imbalance. The performance of K-means SMOTE is compared to three popular oversampling methods (Random Oversampling, SMOTE and Borderline-SMOTE) using seven remote sensing benchmark datasets, three classifiers (Logistic Regression, K-Nearest Neighbors and Random Forest Classifier) and three evaluation metrics using a five-fold cross-validation approach with three different initialization seeds. The statistical analysis of the results show that the proposed method consistently outperforms the remaining oversamplers producing higher quality land cover classifications. These results suggest that LULC data can benefit significantly from the use of more sophisticated oversamplers as spectral signatures for the same class can vary according to geographical distribution.
Increasing the effectiveness of active learning
Publication . Fonseca, Joao; Douzas, Georgios; Bacao, Fernando; Information Management Research Center (MagIC) - NOVA Information Management School; NOVA Information Management School (NOVA IMS); Molecular Diversity Preservation International (MDPI)
In remote sensing, Active Learning (AL) has become an important technique to collect informative ground truth data “on-demand” for supervised classification tasks. Despite its effectiveness, it is still significantly reliant on user interaction, which makes it both expensive and time consuming to implement. Most of the current literature focuses on the optimization of AL by modifying the selection criteria and the classifiers used. Although improvements in these areas will result in more effective data collection, the use of artificial data sources to reduce human–computer interaction remains unexplored. In this paper, we introduce a new component to the typical AL framework, the data generator, a source of artificial data to reduce the amount of user-labeled data required in AL. The implementation of the proposed AL framework is done using Geometric SMOTE as the data generator. We compare the new AL framework to the original one using similar acquisition functions and classifiers over three AL-specific performance metrics in seven benchmark datasets. We show that this modification of the AL framework significantly reduces cost and time requirements for a successful AL implementation in all of the datasets used in the experiment.
Improved terrain type classification using UAV downwash dynamic texture effect
Publication . Carvalho, João Pedro Leal Abalada de Matos; Fonseca, José; Mora, André
The ability to autonomously navigate in an unknown, dynamic environment, while at
the same time classifying various terrain types, are significant challenges still faced by
the computer vision research community. Addressing these problems is of great interest
for the development of collaborative autonomous navigation robots. For example, an
Unmanned Aerial Vehicle (UAV) can be used to determine a path, while an Unmanned
Surface Vehicle (USV) follows that path to reach the target destination. For the UAV to be
able to determine if a path is valid or not, it must be able to identify the type of terrain it
is flying over. With the help of its rotor air flow (known as downwash e↵ect), it becomes
possible to extract advanced texture features, used for terrain type classification.
This dissertation presents a complete analysis on the extraction of static and dynamic
texture features, proposing various algorithms and analyzing their pros and cons. A
UAV equipped with a single RGB camera was used to capture images and a Multilayer
Neural Network was used for the automatic classification of water and non-water-type
terrains by means of the downwash e↵ect created by the UAV rotors. The terrain type
classification results are then merged into a georeferenced dynamic map, where it is
possible to distinguish between water and non-water areas in real time.
To improve the algorithms’ processing time, several sequential processes were con verted into parallel processes and executed in the UAV onboard GPU with the CUDA
framework achieving speedups up to 10x. A comparison between the processing time
of these two processing modes, sequential in the CPU and parallel in the GPU, is also
presented in this dissertation.
All the algorithms were developed using open-source libraries, and were analyzed
and validated both via simulation and real environments. To evaluate the robustness of
the proposed algorithms, the studied terrains were tested with and without the presence
of the downwash e↵ect. It was concluded that the classifier could be improved by per forming combinations between static and dynamic features, achieving an accuracy higher
than 99% in the classification of water and non-water terrain.
Spatially Stratified and Multi-Stage Approach for National Land Cover Mapping Based on Sentinel-2 Data and Expert Knowledge
Publication . Costa, Hugo; Benevides, Pedro; Moreira, Francisco D.; Moraes, Daniel; Caetano, Mário; NOVA Information Management School (NOVA IMS); Information Management Research Center (MagIC) - NOVA Information Management School; Molecular Diversity Preservation International (MDPI)
Portugal is building a land cover monitoring system to deliver land cover products annually for its mainland territory. This paper presents the methodology developed to produce a prototype relative to 2018 as the first land cover map of the future annual map series (COSsim). A total of thirteen land cover classes are represented, including the most important tree species in Portugal. The mapping approach developed includes two levels of spatial stratification based on landscape dynamics. Strata are analysed independently at the higher level, while nested sublevels can share data and procedures. Multiple stages of analysis are implemented in which subsequent stages improve the outputs of precedent stages. The goal is to adjust mapping to the local landscape and tackle specific problems or divide complex mapping tasks in several parts. Supervised classification of Sentinel-2 time series and post-classification analysis with expert knowledge were performed throughout four stages. The overall accuracy of the map is estimated at 81.3% (±2.1) at the 95% confidence level. Higher thematic accuracy was achieved in southern Portugal, and expert knowledge significantly improved the quality of the map.
Imbalanced learning in land cover classification
Publication . Douzas, Georgios; Bacao, Fernando; Fonseca, Joao; Khudinyan, Manvel; Information Management Research Center (MagIC) - NOVA Information Management School; NOVA Information Management School (NOVA IMS); Molecular Diversity Preservation International (MDPI)
The automatic production of land use/land cover maps continues to be a challenging problem, with important impacts on the ability to promote sustainability and good resource management. The ability to build robust automatic classifiers and produce accurate maps can have a significant impact on the way we manage and optimize natural resources. The difficulty in achieving these results comes from many different factors, such as data quality and uncertainty. In this paper, we address the imbalanced learning problem, a common and difficult conundrum in remote sensing that affects the quality of classification results, by proposing Geometric-SMOTE, a novel oversampling method, as a tool for addressing the imbalanced learning problem in remote sensing. Geometric-SMOTE is a sophisticated oversampling algorithm which increases the quality of the instances generated in previous methods, such as the synthetic minority oversampling technique. The performance of Geometric- SMOTE, in the LUCAS (Land Use/Cover Area Frame Survey) dataset, is compared to other oversamplers using a variety of classifiers. The results show that Geometric-SMOTE significantly outperforms all the other oversamplers and improves the robustness of the classifiers. These results indicate that, when using imbalanced datasets, remote sensing researchers should consider the use of these new generation oversamplers to increase the quality of the classification results.
Unidades organizacionais
Descrição
Palavras-chave
Contribuidores
Financiadores
Entidade financiadora
Fundação para a Ciência e a Tecnologia
Programa de financiamento
3599-PPCDT
Número da atribuição
DSAIPA/AI/0100/2018