Gabriel Henrique de Almeida Pereira
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Gabriel Henrique de Almeida Pereira has a total of 41 co-authors. Here are the top co-authors by number of shared publications:
- Jorge Antonio Silva Centeno
Universidade Federal do Paraná
4 shared publications - Bogdan Tomoyuki Nassu
Graduate Program in Applied Computing (PPGCA), Federal University of TechnologyParan (UTFPR), Curitiba, Paran, Brazil
3 shared publications - Rodrigo Minetto
Universidade Tecnológica Federal do Paraná (UTFPR) -Curitiba Brasil
3 shared publications - André Minoro Fusioka
Department of Informatics, Federal University of TechnologyParana (UTFPR), Curitiba, Brazil
2 shared publications - Clóvis Cechim Júnior
Universidade Federal do Paraná, Brazil
2 shared publications - João Vitor Meza Bravo
Universidade Federal de Uberlândia, Instituto de Geografia, Geociências e Saúde Coletiva, Uberlândia, MG, Brasil
2 shared publications
Latest Publications
- Open Access
- 1 January 2026
- Published by FapUNIFESP (SciELO) in Revista Brasileira de Engenharia Agrícola E Ambiental
- Vol. 30 (5)
Use of harmonic time series to detect deforestation in tropical native forests with Sentinel-2 imagery
The rapid expansion of deforested areas in tropical forests, combined with the lack of automated, scalable monitoring methods with high temporal and spatial resolution, underscores the need for efficient systems capable of early detection of changes in forest cover. This study aimed to develop a native forest deforestation detection system based on reference forest mapping using Sentinel-2 images with 10 m spatial resolution, and to implement it on the Google Earth Engine platform. The proposed methodology consists of applying harmonic analysis to the same region of the same forest fragment, using a reference map of the native forest for comparison with more recent images, free of clouds and atmospherically corrected. The system showed high performance, detecting an average of 7.85 deforestation events per day and achieving an overall accuracy of 97.5%, demonstrating its ability to identify subtle vegetation disturbances. In conclusion, the proposed methodology represents an operational advancement for near-real-time monitoring, providing reliable support for enforcement actions and environmental conservation.
- 1 January 2026
- Published by FapUNIFESP (SciELO) in Revista Brasileira de Engenharia Agrícola E Ambiental
- Vol. 30 (5)
Open AccessUse of harmonic time series to detect deforestation in tropical native forests with Sentinel-2 imagery
The rapid expansion of deforested areas in tropical forests, combined with the lack of automated, scalable monitoring methods with high temporal and spatial resolution, underscores the need for efficient systems capable of early detection of changes in forest cover. This study aimed to develop a native forest deforestation detection system based on reference forest mapping using Sentinel-2 images with 10 m spatial resolution, and to implement it on the Google Earth Engine platform. The proposed methodology consists of applying harmonic analysis to the same region of the same forest fragment, using a reference map of the native forest for comparison with more recent images, free of clouds and atmospherically corrected. The system showed high performance, detecting an average of 7.85 deforestation events per day and achieving an overall accuracy of 97.5%, demonstrating its ability to identify subtle vegetation disturbances. In conclusion, the proposed methodology represents an operational advancement for near-real-time monitoring, providing reliable support for enforcement actions and environmental conservation.
- Open Access
A Method for Mapping and Associating Burned Areas with Agricultural Practices Within the Brazilian Cerrado
Fire occurs naturally and anthropogenically in the Cerrado biome, influenced by hydrology, climate, topography, and land use. Mapping burned areas is essential for understanding the causes of fire and improving prevention and regulation. However, fire scars are often confused with bare soil in agricultural regions. This study presents a method for mapping burned areas using spectral indices and artificial neural networks (ANN). We evaluated the accuracy of these techniques and identified the best input variables for scar detection. Using Sentinel-2 images from 2018 to 2021 during dry periods, we applied NDVI, SAVI, NBR, and CSI indices. The study included two stages: first, finding optimal classification configurations for fire scars, and second, mapping land use and cover with fire scars and crops. Results showed that using all Sentinel-2 bands and the four indices post-fire achieved over 93.7% accuracy and a kappa index of 0.92. Fire scars were mainly located in areas with temporary crops like soybean, sugarcane, rice, and cotton. This low-cost method allows for effective monitoring of fire scars, underscoring the need to regulate agricultural practices in the Cerrado, where burning poses environmental and health risks.
Open AccessA Method for Mapping and Associating Burned Areas with Agricultural Practices Within the Brazilian Cerrado
Fire occurs naturally and anthropogenically in the Cerrado biome, influenced by hydrology, climate, topography, and land use. Mapping burned areas is essential for understanding the causes of fire and improving prevention and regulation. However, fire scars are often confused with bare soil in agricultural regions. This study presents a method for mapping burned areas using spectral indices and artificial neural networks (ANN). We evaluated the accuracy of these techniques and identified the best input variables for scar detection. Using Sentinel-2 images from 2018 to 2021 during dry periods, we applied NDVI, SAVI, NBR, and CSI indices. The study included two stages: first, finding optimal classification configurations for fire scars, and second, mapping land use and cover with fire scars and crops. Results showed that using all Sentinel-2 bands and the four indices post-fire achieved over 93.7% accuracy and a kappa index of 0.92. Fire scars were mainly located in areas with temporary crops like soybean, sugarcane, rice, and cotton. This low-cost method allows for effective monitoring of fire scars, underscoring the need to regulate agricultural practices in the Cerrado, where burning poses environmental and health risks.
- 14 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 21 (1545598X) , 1-5
Sentinel-2 Active Fire Segmentation: Analyzing Convolutional and Transformer Architectures, Knowledge Transfer, Fine-Tuning, and Seam Lines
Active fire segmentation in satellite imagery is a critical remote sensing task, providing essential support for planning, decision-making, and policy development. Several techniques have been proposed for this problem over the years, generally based on specific equations and thresholds, which are sometimes empirically chosen. Some satellites, such as MODIS and Landsat-8 have consolidated algorithms for this task. However, for other important satellites such as Sentinel-2, this is still an open problem. In this paper, we explore the possibility of using transfer learning to train convolutional and transformer-based deep architectures (U-Net, DeepLabV3+ and SegFormer) for active fire segmentation. We pre-train these architectures based on Landsat-8 images and automatically labeled samples, and fine-tune them to Sentinel-2 images. Experiments show that the proposed method achieves F 1-scores of up to 88.4% for Sentinel-2 images, outperforming three threshold-based algorithms by at least 19%, while maintaining a low demand for manually labeled samples. We also address detection over seam-line regions that present a particular challenge for existing methods. The source code and trained models are available on GitHub.
- 14 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 21 (1545598X) , 1-5
Sentinel-2 Active Fire Segmentation: Analyzing Convolutional and Transformer Architectures, Knowledge Transfer, Fine-Tuning, and Seam Lines
Active fire segmentation in satellite imagery is a critical remote sensing task, providing essential support for planning, decision-making, and policy development. Several techniques have been proposed for this problem over the years, generally based on specific equations and thresholds, which are sometimes empirically chosen. Some satellites, such as MODIS and Landsat-8 have consolidated algorithms for this task. However, for other important satellites such as Sentinel-2, this is still an open problem. In this paper, we explore the possibility of using transfer learning to train convolutional and transformer-based deep architectures (U-Net, DeepLabV3+ and SegFormer) for active fire segmentation. We pre-train these architectures based on Landsat-8 images and automatically labeled samples, and fine-tune them to Sentinel-2 images. Experiments show that the proposed method achieves F 1-scores of up to 88.4% for Sentinel-2 images, outperforming three threshold-based algorithms by at least 19%, while maintaining a low demand for manually labeled samples. We also address detection over seam-line regions that present a particular challenge for existing methods. The source code and trained models are available on GitHub.
- Open Access
- 2 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 17 (19391404) , 14093-14108
Active Fire Segmentation: A Transfer Learning Study From Landsat-8 to Sentinel-2
Active fire segmentation through satellite imagery is fundamental to prevention and damage prediction. Algorithms to address this problem usually rely on sensor-specific thresholds, empirically chosen based on a few image samples, and thus, are susceptible to many errors. Deep learning algorithms automatically extract information through various levels of abstraction, avoiding human intervention for feature engineering. However, training a deep network from scratch requires massive labeled data, and efforts in this direction have already been made for the Landsat-8 satellite. For Sentinel-2, another important satellite that has band wavelength similarities with Landsat-8, there is a limited number of such initiatives, with few studies even concerning hand-crafted (traditional) algorithms. In this context, we explored in this paper the transfer of knowledge for active fire segmentation from Landsat-8 to Sentinel-2, avoiding the need of a vast amount of labeled data from Sentinel-2, and also reducing the computational resources for training. We also compiled a benchmark containing 12,584 image patches extracted from 26 Sentinel-2 images from around the globe, along with manually annotated fire pixels, to assess the algorithm's response compared to a human specialist. In a series of transfer learning experiments by using the U-Net architecture, we showed that even a single labeled image from Sentinel-2 for training was sufficient to allow achieving an $F$ -score metric of 84.9% when Landsat-8 knowledge is transferred and further fine-tuned in the machine learning process, while the best hand-crafted algorithm designed for Sentinel-2 achieved an $F$ -score of 75.8% in the segmentation task. This project is available on GitHub.
- 2 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 17 (19391404) , 14093-14108
Open AccessActive Fire Segmentation: A Transfer Learning Study From Landsat-8 to Sentinel-2
Active fire segmentation through satellite imagery is fundamental to prevention and damage prediction. Algorithms to address this problem usually rely on sensor-specific thresholds, empirically chosen based on a few image samples, and thus, are susceptible to many errors. Deep learning algorithms automatically extract information through various levels of abstraction, avoiding human intervention for feature engineering. However, training a deep network from scratch requires massive labeled data, and efforts in this direction have already been made for the Landsat-8 satellite. For Sentinel-2, another important satellite that has band wavelength similarities with Landsat-8, there is a limited number of such initiatives, with few studies even concerning hand-crafted (traditional) algorithms. In this context, we explored in this paper the transfer of knowledge for active fire segmentation from Landsat-8 to Sentinel-2, avoiding the need of a vast amount of labeled data from Sentinel-2, and also reducing the computational resources for training. We also compiled a benchmark containing 12,584 image patches extracted from 26 Sentinel-2 images from around the globe, along with manually annotated fire pixels, to assess the algorithm's response compared to a human specialist. In a series of transfer learning experiments by using the U-Net architecture, we showed that even a single labeled image from Sentinel-2 for training was sufficient to allow achieving an $F$ -score metric of 84.9% when Landsat-8 knowledge is transferred and further fine-tuned in the machine learning process, while the best hand-crafted algorithm designed for Sentinel-2 achieved an $F$ -score of 75.8% in the segmentation task. This project is available on GitHub.
- Open Access
- 1 January 2023
- Published by FapUNIFESP (SciELO) in RBRH
- Vol. 28, e29
A Web-GIS for decision making to achieve water quality standards of water bodies through collaborative watershed modeling
Ensuring compliance with the minimum Water Quality Standards stipulated by law demands the implementation of strategic measures within the watershed. Water pollution modeling serves as a tool to guide the formulations of effective pollution control strategies. However, the inherent complexity of calibration, spatial-temporal variability, and uncertainty, in addition to effective communication of technical information to decision makers, makes it challenging to prioritize actions, implement them, and allocate resources efficiently. This paper presents the implementation of a Web-GIS for decision making support which combines collaborative hydrological and hydrodynamic modeling results with an interactive visualization of the Paranapanema river basin in the South of Brazil. The Web-GIS aimed to overcome the difficulty of presenting scientific results to decision-makers, gathering and harmonizing diverse datasets. Combining information from the watershed, the pollutant loads estimations for three substances (phosphorus, nitrogen, and BOD), the resulting concentrations in rivers and reservoirs, as well as the results for different future scenarios into a unified platform. It is expected that decision-making regarding water bodies framework will be facilitated by identifying the primary sources and pathways of pollution, prioritizing basins with the highest load production, an determining realistic possibilities of load reduction through effective measures.
- 1 January 2023
- Published by FapUNIFESP (SciELO) in RBRH
- Vol. 28, e29
Open AccessA Web-GIS for decision making to achieve water quality standards of water bodies through collaborative watershed modeling
Ensuring compliance with the minimum Water Quality Standards stipulated by law demands the implementation of strategic measures within the watershed. Water pollution modeling serves as a tool to guide the formulations of effective pollution control strategies. However, the inherent complexity of calibration, spatial-temporal variability, and uncertainty, in addition to effective communication of technical information to decision makers, makes it challenging to prioritize actions, implement them, and allocate resources efficiently. This paper presents the implementation of a Web-GIS for decision making support which combines collaborative hydrological and hydrodynamic modeling results with an interactive visualization of the Paranapanema river basin in the South of Brazil. The Web-GIS aimed to overcome the difficulty of presenting scientific results to decision-makers, gathering and harmonizing diverse datasets. Combining information from the watershed, the pollutant loads estimations for three substances (phosphorus, nitrogen, and BOD), the resulting concentrations in rivers and reservoirs, as well as the results for different future scenarios into a unified platform. It is expected that decision-making regarding water bodies framework will be facilitated by identifying the primary sources and pathways of pollution, prioritizing basins with the highest load production, an determining realistic possibilities of load reduction through effective measures.
- Open Access
- 28 August 2019
- Published by Universidade Federal do Parana in Raega - O Espaço Geográfico em Análise
- Vol. 46 (3) , 88-100
MULTITEMPORAL ANALYSIS OF SAR IMAGES FOR DETECTION OF FLOODED AREAS IN PANTANAL
MULTITEMPORAL ANALYSIS OF SAR IMAGES FOR DETECTION OF FLOODED AREAS IN PANTANAL
- 28 August 2019
- Published by Universidade Federal do Parana in Raega - O Espaço Geográfico em Análise
- Vol. 46 (3) , 88-100
Open AccessMULTITEMPORAL ANALYSIS OF SAR IMAGES FOR DETECTION OF FLOODED AREAS IN PANTANAL
MULTITEMPORAL ANALYSIS OF SAR IMAGES FOR DETECTION OF FLOODED AREAS IN PANTANAL
- 2 January 2019
- Published by Taylor & Francis in Canadian Journal of Remote Sensing
- Vol. 45 (1) , 16-25
Artificial Neural Networks and Data Mining Techniques for Summer Crop Discrimination: A New Approach
The objective of this research was to distinguish and estimate cultivated areas of soybean and corn in Paraná State, Brazil, in the 2014/2015 crop season. The main obstacle in mapping summer crops using vegetation index images is to separate the cultivated areas with soybean and corn. These crops planted in a similar period present similar spectral signatures. Thus, with the use of Data Mining techniques (DM) and Artificial Neural Network (ANN) it was possible to carry out the crop mapping, even for those that present similarities in spectral-temporal profile of vegetation indexes. The accuracy obtained in the mappings resulted in a KI (Kappa Index) of 0.78 and 89% of OA (overall accuracy) indicating a high accuracy in the separation of soybean and corn summer crops.
- 2 January 2019
- Published by Taylor & Francis in Canadian Journal of Remote Sensing
- Vol. 45 (1) , 16-25
Artificial Neural Networks and Data Mining Techniques for Summer Crop Discrimination: A New Approach
The objective of this research was to distinguish and estimate cultivated areas of soybean and corn in Paraná State, Brazil, in the 2014/2015 crop season. The main obstacle in mapping summer crops using vegetation index images is to separate the cultivated areas with soybean and corn. These crops planted in a similar period present similar spectral signatures. Thus, with the use of Data Mining techniques (DM) and Artificial Neural Network (ANN) it was possible to carry out the crop mapping, even for those that present similarities in spectral-temporal profile of vegetation indexes. The accuracy obtained in the mappings resulted in a KI (Kappa Index) of 0.78 and 89% of OA (overall accuracy) indicating a high accuracy in the separation of soybean and corn summer crops.
- Open Access
- 17 August 2018
- Published by Portal de Periódicos UFU - Universidade Federal de Uberlandia in Revista Brasileira de Cartografia
- Vol. 69 (8)
O USO DE REALIDADE AUMENTADA PARA MAPAS OFICIAIS PARA FINS DE PLANEJAMENTO E GESTÃO
Interfaces de sistemas computacionais cada vez mais naturais auxiliam na redução da barreira ao acesso a elementos da informática e, simultaneamente, aumentando o poder do usuário, auxiliando-o na execução de suas tarefas. Estas interfaces computacionais mais naturais podem possibilitar novas formas de interação com a informação assim como com o ambiente que o circunda. Dentre os sistemas que prometem auxiliar os usuários em suas tarefas, inclusive de leitura de mapa, os baseados em Realidade Aumentada mostram-se bastante promissores. Entretanto, pouco deste tema tem sido abordado, principalmente, no Brasil. Por esse motivo, elaborou-se a presente pesquisa na intenção de desenvolver um protótipo de Realidade Aumentada para interação com mapas e dados geográficos. Foram utilizados dados geoespaciais referentes às características físicas, políticas, social e demográfica, econômicas e ambientais, integradas sobre um mapa político rodoviário oficial do Estado do Paraná. É possível destacar que o sistema de Realidade Aumentada desenvolvido traz a capacidade de manipulação de dados geoespaciais digitais sobre um mapa impresso, trazendo novas oportunidades de interação com o mapa ou dados geoespaciais digitais, além da possibilidade de exploração de vários formatos de apresentação dos dados, como camadas 2D, 3D e animações. Desta forma, aplicações de Realidade Aumentada como a descrita neste trabalho se tornam importantes ferramentas no auxílio às questões de planejamento, gestão e ordenamento do território, uma vez que a integração dos dados pode auxiliar no processo de entendimento e compreensão dos espaço geográfico por parte dos seus usuários.
- 17 August 2018
- Published by Portal de Periódicos UFU - Universidade Federal de Uberlandia in Revista Brasileira de Cartografia
- Vol. 69 (8)
Open AccessO USO DE REALIDADE AUMENTADA PARA MAPAS OFICIAIS PARA FINS DE PLANEJAMENTO E GESTÃO
Interfaces de sistemas computacionais cada vez mais naturais auxiliam na redução da barreira ao acesso a elementos da informática e, simultaneamente, aumentando o poder do usuário, auxiliando-o na execução de suas tarefas. Estas interfaces computacionais mais naturais podem possibilitar novas formas de interação com a informação assim como com o ambiente que o circunda. Dentre os sistemas que prometem auxiliar os usuários em suas tarefas, inclusive de leitura de mapa, os baseados em Realidade Aumentada mostram-se bastante promissores. Entretanto, pouco deste tema tem sido abordado, principalmente, no Brasil. Por esse motivo, elaborou-se a presente pesquisa na intenção de desenvolver um protótipo de Realidade Aumentada para interação com mapas e dados geográficos. Foram utilizados dados geoespaciais referentes às características físicas, políticas, social e demográfica, econômicas e ambientais, integradas sobre um mapa político rodoviário oficial do Estado do Paraná. É possível destacar que o sistema de Realidade Aumentada desenvolvido traz a capacidade de manipulação de dados geoespaciais digitais sobre um mapa impresso, trazendo novas oportunidades de interação com o mapa ou dados geoespaciais digitais, além da possibilidade de exploração de vários formatos de apresentação dos dados, como camadas 2D, 3D e animações. Desta forma, aplicações de Realidade Aumentada como a descrita neste trabalho se tornam importantes ferramentas no auxílio às questões de planejamento, gestão e ordenamento do território, uma vez que a integração dos dados pode auxiliar no processo de entendimento e compreensão dos espaço geográfico por parte dos seus usuários.
- Open Access
- 1 August 2018
- Published by MDPI AG in Big Data and Cognitive Computing
- Vol. 2 (3) , 20
A User Study of a Prototype of a Spatial Augmented Reality System for Education and Interaction with Geographic Data
Recent technological advancements in many areas have changed the way that individuals interact with the world. Some daily tasks require visualization skills, especially when in a map-reading context. Augmented Reality systems could provide substantial improvement to geovisualization once it enhances a real scene with virtual information. However, relatively little research has worked on assessing the effective contribution of such systems during map reading. So, this research aims to provide a first look into the usability of an Augmented Reality system prototype for interaction with geoinformation. For this purpose, we have designed an activity with volunteers in order to assess the system prototype usability. We have interviewed 14 users (three experts and 11 non-experts), where experts were subjects with the following characteristics: a professor; with a PhD degree in Cartography, GIS, Geography, or Environmental Sciences/Water Resources; and with experience treating spatial information related to water resources. The activity aimed to detect where the system really helps the user to interpret a hydrographic map and how the users were helped by the Augmented Reality system prototype. We may conclude that the Augmented Reality system was helpful to the users during the map reading, as well as allowing the construction of spatial knowledge within the proposed scenario.
- 1 August 2018
- Published by MDPI AG in Big Data and Cognitive Computing
- Vol. 2 (3) , 20
Open AccessA User Study of a Prototype of a Spatial Augmented Reality System for Education and Interaction with Geographic Data
Recent technological advancements in many areas have changed the way that individuals interact with the world. Some daily tasks require visualization skills, especially when in a map-reading context. Augmented Reality systems could provide substantial improvement to geovisualization once it enhances a real scene with virtual information. However, relatively little research has worked on assessing the effective contribution of such systems during map reading. So, this research aims to provide a first look into the usability of an Augmented Reality system prototype for interaction with geoinformation. For this purpose, we have designed an activity with volunteers in order to assess the system prototype usability. We have interviewed 14 users (three experts and 11 non-experts), where experts were subjects with the following characteristics: a professor; with a PhD degree in Cartography, GIS, Geography, or Environmental Sciences/Water Resources; and with experience treating spatial information related to water resources. The activity aimed to detect where the system really helps the user to interpret a hydrographic map and how the users were helped by the Augmented Reality system prototype. We may conclude that the Augmented Reality system was helpful to the users during the map reading, as well as allowing the construction of spatial knowledge within the proposed scenario.