Deep Transfer Learning for Land Use and Land Cover Classification: A Comparative Study

Development of a convolutional neural network to accurately detect land use and land cover

The detection and modeling of Land Use and Land Cover (LULC) play pivotal roles in natural resource management, environmental modeling and assessment, and ecological connectivity management. However, addressing LULCC detection and modeling constitutes a complex data-driven process. In the present study, a Convolutional Neural Network (CNN) is employed due to its great potential in image classification. The development of these tools applies the deep learning method. A methodology has been developed that classifies the set of land uses in a natural area of special protection. This study area covers the Sierra del Cando (Galicia, northwest Spain), considered by the European Union as a Site of Community Interest and integrated in the Natura 2000 Network. The results of the CNN model developed show an accuracy of 91 % on training dataset and 88 % on test dataset. In addition, the model was tested on images of the study area, both from Sentinel-2 and PNOA. Despite some confusion especially in the residential class due to the characteristics in this area, CNNs prove to be a powerful classification tool.•Classifications based on a CNN model•LULC are classified into 10 different classes•Training and test accuracy are 91 % and 88 %, respectively.

Generation and classification of patch-based land use and land cover dataset in diverse Indian landscapes: a comparative study of machine learning and deep learning models

In the context of environmental and social applications, the analysis of land use and land cover (LULC) holds immense significance. The growing accessibility of remote sensing (RS) data has led to the development of LULC benchmark datasets, especially pivotal for intricate image classification tasks. This study addresses the scarcity of such benchmark datasets across diverse settings, with a particular focus on the distinctive landscape of India. The study entails the creation of patch-based datasets, consisting of 4000 labelled images spanning four distinct LULC classes derived from Sentinel-2 satellite imagery. For the subsequent classification task, three traditional machine learning (ML) models and three convolutional neural networks (CNNs) were employed. Despite facing several challenges throughout the process of dataset generation and subsequent classification, the CNN models consistently attained an overall accuracy of 90% or more. Notably, one of the ML models stood out with 96% accuracy, surpassing CNNs in this specific context. The study also conducts a comparative analysis of ML models on existing benchmark datasets, revealing higher prediction accuracy when dealing with fewer LULC classes. Thus, the selection of an appropriate model hinges on the given task, available resources, and the necessary trade-offs between performance and efficiency, particularly crucial in resource-constrained settings. The standardized benchmark dataset contributes valuable insights into the relative performance of deep CNN and ML models in LULC classification, providing a comprehensive understanding of their strengths and weaknesses.

CloudDenseNet: Lightweight Ground-Based Cloud Classification Method for Large-Scale Datasets Based on Reconstructed DenseNet

Cloud observation serves as the fundamental bedrock for acquiring comprehensive cloud-related information. The categorization of distinct ground-based clouds holds profound implications within the meteorological domain, boasting significant applications. Deep learning has substantially improved ground-based cloud classification, with automated feature extraction being simpler and far more accurate than using traditional methods. A reengineering of the DenseNet architecture has given rise to an innovative cloud classification method denoted as CloudDenseNet. A novel CloudDense Block has been meticulously crafted to amplify channel attention and elevate the salient features pertinent to cloud classification endeavors. The lightweight CloudDenseNet structure is designed meticulously according to the distinctive characteristics of ground-based clouds and the intricacies of large-scale diverse datasets, which amplifies the generalization ability and elevates the recognition accuracy of the network. The optimal parameter is obtained by combining transfer learning with designed numerous experiments, which significantly enhances the network training efficiency and expedites the process. The methodology achieves an impressive 93.43% accuracy on the large-scale diverse dataset, surpassing numerous published methods. This attests to the substantial potential of the CloudDenseNet architecture for integration into ground-based cloud classification tasks.

CRABR-Net: A Contextual Relational Attention-Based Recognition Network for Remote Sensing Scene Objective

Remote sensing scene objective recognition (RSSOR) plays a serious application value in both military and civilian fields. Convolutional neural networks (CNNs) have greatly enhanced the improvement of intelligent objective recognition technology for remote sensing scenes, but most of the methods using CNN for high-resolution RSSOR either use only the feature map of the last layer or directly fuse the feature maps from various layers in the "summation" way, which not only ignores the favorable relationship information between adjacent layers but also leads to redundancy and loss of feature map, which hinders the improvement of recognition accuracy. In this study, a contextual, relational attention-based recognition network (CRABR-Net) was presented, which extracts different convolutional feature maps from CNN, focuses important feature content by using a simple, parameter-free attention module (SimAM), fuses the adjacent feature maps by using the complementary relationship feature map calculation, improves the feature learning ability by using the enhanced relationship feature map calculation, and finally uses the concatenated feature maps from different layers for RSSOR. Experimental results show that CRABR-Net exploits the relationship between the different CNN layers to improve recognition performance, achieves better results compared to several state-of-the-art algorithms, and the average accuracy on AID, UC-Merced, and RSSCN7 can be up to 96.46%, 99.20%, and 95.43% with generic training ratios.

Image classification adversarial attack with improved resizing transformation and ensemble models

Convolutional neural networks have achieved great success in computer vision, but incorrect predictions would be output when applying intended perturbations on original input. These human-indistinguishable replicas are called adversarial examples, which on this feature can be used to evaluate network robustness and security. White-box attack success rate is considerable, when already knowing network structure and parameters. But in a black-box attack, the adversarial examples success rate is relatively low and the transferability remains to be improved. This article refers to model augmentation which is derived from data augmentation in training generalizable neural networks, and proposes resizing invariance method. The proposed method introduces improved resizing transformation to achieve model augmentation. In addition, ensemble models are used to generate more transferable adversarial examples. Extensive experiments verify the better performance of this method in comparison to other baseline methods including the original model augmentation method, and the black-box attack success rate is improved on both the normal models and defense models.

Remote sensing identification of green plastic cover in urban built-up areas

Urban renewal can transform areas that are not adapted to modern urban life, allowing them to redevelop and flourish; however, the renewal process generates many new construction sites, producing environmentally harmful construction dust. The widespread use of urban green plastic cover (GPC) at construction sites and the development of high-resolution satellites have made it possible to extract the spatial distribution of construction sites and provide a basis for environmental protection authorities to protect against dust sources. Existing GPC extraction methods based on remote sensing images are either difficult to obtain the exact boundary of GPC or cannot provide corresponding algorithms according to different application scenarios. In order to determine the distribution of green plastic cover in the built-up area, this paper selects a variety of typical machine learning algorithms to classify the land cover of the test area image and selects K-nearest neighbor as the best machine learning algorithm through accuracy evaluation. Then multiple deep learning methods were used and the top networks with high overall scores were selected by comparing various aspects. Then these networks were used to predict the GPC of the test area image, and the accuracy evaluation results showed that the segmentation accuracy of deep learning was much higher than that of machine learning methods, but it took more time to predict. Therefore, combining different application scenarios, this paper gives the corresponding suggested methods for GPC extraction.

Inception Convolution and Feature Fusion for Person Search

With the rapid advancement of deep learning theory and hardware device computing capacity, computer vision tasks, such as object detection and instance segmentation, have entered a revolutionary phase in recent years. As a result, extremely challenging integrated tasks, such as person search, might develop quickly. The majority of efficient network frameworks, such as Seq-Net, are based on Faster R-CNN. However, because of the parallel structure of Faster R-CNN, the performance of re-ID can be significantly impacted by the single-layer, low resolution, and occasionally overlooked check feature diagrams retrieved during pedestrian detection. To address these issues, this paper proposed a person search methodology based on an inception convolution and feature fusion module (IC-FFM) using Seq-Net (Sequential End-to-end Network) as the benchmark. First, we replaced the general convolution in ResNet-50 with the new inception convolution module (ICM), allowing the convolution operation to effectively and dynamically distribute various channels. Then, to improve the accuracy of information extraction, the feature fusion module (FFM) was created to combine multi-level information using various levels of convolution. Finally, Bounding Box regression was created using convolution and the double-head module (DHM), which considerably enhanced the accuracy of pedestrian retrieval by combining global and fine-grained information. Experiments on CHUK-SYSU and PRW datasets showed that our method has higher accuracy than Seq-Net. In addition, our method is simpler and can be easily integrated into existing two-stage frameworks.

Research on land cover type classification method based on improved MaskFormer for remote sensing images

High-resolution remote sensing images have the characteristics of wide imaging coverage, rich spectral information and unobstructed by terrain and features. All of them provide convenient conditions for people to study land cover types. However, most existing remote sensing image land cover datasets are only labeled with some remote sensing images of low elevation plain areas, which is highly different from the topography and landscape of highland mountainous areas. In this study, we construct a Qilian County grassland ecological element dataset to provide data support for highland ecological protection. To highlight the characteristics of vegetation, our dataset only includes the RGB spectrum fused with the near-infrared spectrum. We then propose a segmentation network, namely, the Shunted-MaskFormer network, by using a mask-based classification method, a multi-scale, high-efficiency feature extraction module and a data-dependent upsampling method. The extraction of grassland land types from 2 m resolution remote sensing images in Qilian County was completed, and the generalization ability of the model on a small Gaofen Image Dataset (GID) verified. Results: (1) The MIoU of the optimised network model in the Qilian grassland dataset reached 80.75%, which is 2.37% higher compared to the suboptimal results; (2) the optimized network model achieves better segmentation results even for small sample classes in data sets with unbalanced sample distribution; (3) the highest MIOU of 72.3% is achieved in the GID dataset of open remote sensing images containing five categories; (4) the size of the optimized model is only one-third of the sub-optimal model.

Land Use Land Cover Labeling of GLOBE Images Using a Deep Learning Fusion Model

Most of the land use land cover classification methods presented in the literature have been conducted using satellite remote sensing images. High-resolution aerial imagery is now being used for land cover classification. The Global Learning and Observations to Benefit, the Environment land cover image database, is created by citizen scientists worldwide who use their handheld cameras to take a set of six images per land cover site. These images have clutter due to man-made objects, and the pixel uncertainties result in incorrect labels. The problem of accurate labeling of these land cover images is addressed. An integrated architecture that combines Unet and DeepLabV3 for initial segmentation, followed by a weighted fusion model that combines the segmentation labels, is presented. The land cover images with labels are used for training the deep learning models. The fusion model combines the labels of five images taken from the north, south, east, west, and down directions to assign a unique label to the image sets. 2916 GLOBE images have been labeled with land cover classes using the integrated model with minimal human-in-the-loop annotation. The validation step shows that our architecture of labeling the images results in 90.97% label accuracy. Our fusion model can be used for labeling large databases of land cover classes from RGB images.

A Systematic Evaluation of Ensemble Learning Methods for Fine-Grained Semantic Segmentation of Tuberculosis-Consistent Lesions in Chest Radiographs

Automated segmentation of tuberculosis (TB)-consistent lesions in chest X-rays (CXRs) using deep learning (DL) methods can help reduce radiologist effort, supplement clinical decision-making, and potentially result in improved patient treatment. The majority of works in the literature discuss training automatic segmentation models using coarse bounding box annotations. However, the granularity of the bounding box annotation could result in the inclusion of a considerable fraction of false positives and negatives at the pixel level that may adversely impact overall semantic segmentation performance. This study evaluates the benefits of using fine-grained annotations of TB-consistent lesions toward training the variants of U-Net models and constructing their ensembles for semantically segmenting TB-consistent lesions in both original and bone-suppressed frontal CXRs. The segmentation performance is evaluated using several ensemble methods such as bitwise- AND, bitwise-OR, bitwise-MAX, and stacking. Extensive empirical evaluations showcased that the stacking ensemble demonstrated superior segmentation performance (Dice score: 0.5743, 95% confidence interval: (0.4055, 0.7431)) compared to the individual constituent models and other ensemble methods. To the best of our knowledge, this is the first study to apply ensemble learning to improve fine-grained TB-consistent lesion segmentation performance.

Topology Optimisation under Uncertainties with Neural Networks

Topology optimisation is a mathematical approach relevant to different engineering problems where the distribution of material in a defined domain is distributed in some optimal way, subject to a predefined cost function representing desired (e.g., mechanical) properties and constraints. The computation of such an optimal distribution depends on the numerical solution of some physical model (in our case linear elasticity) and robustness is achieved by introducing uncertainties into the model data, namely the forces acting on the structure and variations of the material stiffness, rendering the task high-dimensional and computationally expensive. To alleviate this computational burden, we develop two neural network architectures (NN) that are capable of predicting the gradient step of the optimisation procedure. Since state-of-the-art methods use adaptive mesh refinement, the neural networks are designed to use a sufficiently fine reference mesh such that only one training phase of the neural network suffices. As a first architecture, a convolutional neural network is adapted to the task. To include sequential information of the optimisation process, a recurrent neural network is constructed as a second architecture. A common 2D bridge benchmark is used to illustrate the performance of the proposed architectures. It is observed that the NN prediction of the gradient step clearly outperforms the classical optimisation method, in particular since larger iteration steps become viable.

Urban Land Use and Land Cover Change Analysis Using Random Forest Classification of Landsat Time Series

Efficient implementation of remote sensing image classification can facilitate the extraction of spatiotemporal information for land use and land cover (LULC) classification. Mapping LULC change can pave the way to investigate the impacts of different socioeconomic and environmental factors on the Earth’s surface. This study presents an algorithm that uses Landsat time-series data to analyze LULC change. We applied the Random Forest (RF) classifier, a robust classification method, in the Google Earth Engine (GEE) using imagery from Landsat 5, 7, and 8 as inputs for the 1985 to 2019 period. We also explored the performance of the pan-sharpening algorithm on Landsat bands besides the impact of different image compositions to produce a high-quality LULC map. We used a statistical pan-sharpening algorithm to increase multispectral Landsat bands’ (Landsat 7–9) spatial resolution from 30 m to 15 m. In addition, we checked the impact of different image compositions based on several spectral indices and other auxiliary data such as digital elevation model (DEM) and land surface temperature (LST) on final classification accuracy based on several spectral indices and other auxiliary data on final classification accuracy. We compared the classification result of our proposed method and the Copernicus Global Land Cover Layers (CGLCL) map to verify the algorithm. The results show that: (1) Using pan-sharpened top-of-atmosphere (TOA) Landsat products can produce more accurate results for classification instead of using surface reflectance (SR) alone; (2) LST and DEM are essential features in classification, and using them can increase final accuracy; (3) the proposed algorithm produced higher accuracy (94.438% overall accuracy (OA), 0.93 for Kappa, and 0.93 for F1-score) than CGLCL map (84.4% OA, 0.79 for Kappa, and 0.50 for F1-score) in 2019; (4) the total agreement between the classification results and the test data exceeds 90% (93.37–97.6%), 0.9 (0.91–0.96), and 0.85 (0.86–0.95) for OA, Kappa values, and F1-score, respectively, which is acceptable in both overall and Kappa accuracy. Moreover, we provide a code repository that allows classifying Landsat 4, 5, 7, and 8 within GEE. This method can be quickly and easily applied to other regions of interest for LULC mapping.

An Ecosystem Services-Centric Land Use and Land Cover Classification for a Subbasin of the Tampa Bay Watershed

Land-use and land-cover (LULC) change is a primary driver of terrestrial carbon release, often through the conversion of forest into agriculture or expansion of urban areas. Classification schemes are a key component of landscape analyses. This study creates a novel LULC classification scheme by incorporating ecological data to redefine classes of an existing LULC classification based on variation in above-ground tree carbon. A tree inventory was conducted for 531 plots within a subbasin of the Tampa Bay Watershed, Florida, USA. Above-ground tree carbon was estimated using the i-Tree model. Plots were classified using the Florida Land Use Cover Classification System. Mean quantities of above-ground tree carbon, by class, were tested for statistical differences. A reclassification was conducted based on these differences. Sub-classes within a given “land cover” class were similar for six of the seven classes. Significant differences were found within the “Wetlands” class based on vegetation cover, forming two distinct groups: “Forested Wetlands” and “Non-forested and Mangrove Wetlands”. The urban “land use” class showed differences between “Residential” and “Non-residential” sub-classes, forming two new classes. LULC classifications can sometimes aggregate areas perceived as similar that are in fact distinct regarding ecological variables. These aggregations can obscure the true variation in a parameter at the landscape scale. Therefore, a study’s classification system should be designed to reflect landscape variation in the parameter(s) of interest.

COVID-19 Chest X-ray Classification and Severity Assessment Using Convolutional and Transformer Neural Networks

The coronavirus pandemic started in Wuhan, China in December 2019, and put millions of people in a difficult situation. This fatal virus spread to over 227 countries and the number of infected patients increased to over 400 million cases, causing over 6 million deaths worldwide. Due to the serious consequence of this virus, it is necessary to develop a detection method that can respond quickly to prevent the spreading of COVID-19. Using chest X-ray images to detect COVID-19 is one of the promising techniques; however, with a large number of COVID-19 infected cases every day, the number of radiologists available to diagnose the chest X-ray images is not sufficient. We must have a computer aid system that helps doctors instantly and automatically determine COVID-19 cases. Recently, with the emergence of deep learning methods applied for medical and biomedical uses, using convolutional neural net and transformer applications for chest X-ray images can be a supplement for COVID-19 testing. In this paper, we attempt to classify three types of chest X-ray, which are normal, pneumonia, and COVID-19 using deep learning methods on a customized dataset. We also carry out an experiment on the COVID-19 severity assessment task using a tailored dataset. Five deep learning models were obtained to conduct our experiments: DenseNet121, ResNet50, InceptionNet, Swin Transformer, and Hybrid EfficientNet-DOLG neural networks. The results indicated that chest X-ray and deep learning could be reliable methods for supporting doctors in COVID-19 identification and severity assessment tasks.

A Fault-Line Selection Method for Small-Current Grounded System Based on Deep Transfer Learning

Usually, data-driven methods require many samples and need to train a specific model for each substation instance. As different substation instances have similar fault features, the number of samples required for model training can be significantly reduced if these features are transferred to the substation instances that lack samples. This paper proposes a fault-line selection (FLS) method based on deep transfer learning for small-current grounded systems to solve the problems of unstable training and low FLS accuracy of data-driven methods in small-sample cases. For this purpose, fine-turning and historical averaging techniques are proposed for use in transfer learning to extract similar fault features from other substation instances and transfer these features to target substation instances that lack samples to improve the accuracy and stability of the model. The results show that the proposed method obtains a much higher FLS accuracy than other methods in small-sample cases; it has a strong generalization ability, low misclassification rate, and excellent application value.

Remote Sensing Mapping of Build-Up Land with Noisy Label via Fault-Tolerant Learning

China’s urbanization has dramatically accelerated in recent decades. Land for urban build-up has changed not only in large cities but also in small counties. Land cover mapping is one of the fundamental tasks in the field of remote sensing and has received great attention. However, most current mapping requires a significant manual effort for labeling or classification. It is of great practical value to use the existing low-resolution label data for the classification of higher resolution images. In this regard, this work proposes a method based on noise-label learning for fine-grained mapping of urban build-up land in a county in central China. Specifically, this work produces a build-up land map with a resolution of 10 m based on a land cover map with a resolution of 30 m. Experimental results show that the accuracy of the results is improved by 5.5% compared with that of the baseline method. This notion indicates that the time required to produce a fine land cover map can be significantly reduced using existing coarse-grained data.

Full Convolution Neural Network Combined with Contextual Feature Representation for Cropland Extraction from High-Resolution Remote Sensing Images

The quantity and quality of cropland are the key to ensuring the sustainable development of national agriculture. Remote sensing technology can accurately and timely detect the surface information, and objectively reflect the state and changes of the ground objects. Using high-resolution remote sensing images to accurately extract cropland is the basic task of precision agriculture. The traditional model of cropland semantic segmentation based on the deep learning network is to down-sample high-resolution feature maps to low resolution, and then restore from low-resolution feature maps to high-resolution ideas; that is, obtain low-resolution feature maps through a network, and then recover to high resolution by up-sampling or deconvolution. This will bring about the loss of features, and the segmented image will be more fragmented, without very clear and smooth boundaries. A new methodology for the effective and accurate semantic segmentation cropland of high spatial resolution remote sensing images is presented in this paper. First, a multi-temporal sub-meter cropland sample dataset is automatically constructed based on the prior result data. Then, a fully convolutional neural network combined with contextual feature representation (HRNet-CFR) is improved to complete the extraction of cropland. Finally, the initial semantic segmentation results are optimized by the morphological post-processing approach, and the broken spots are ablated to obtain the internal homogeneous cropland. The proposed method has been validated on the Jilin-1 data and Gaofen Image Dataset (GID) public datasets, and the experimental results demonstrate that it outperforms the state-of-the-art method in cropland extraction accuracy. We selected the comparison of Deeplabv3+ and UPerNet methods in GID. The overall accuracy of our approach is 92.03%, which is 3.4% higher than Deeplabv3+ and 5.12% higher than UperNet.

Land Use and Land Cover Mapping Using Sentinel-2, Landsat-8 Satellite Images, and Google Earth Engine: A Comparison of Two Composition Methods

Accurate and real-time land use/land cover (LULC) maps are important to provide precise information for dynamic monitoring, planning, and management of the Earth. With the advent of cloud computing platforms, time series feature extraction techniques, and machine learning classifiers, new opportunities are arising in more accurate and large-scale LULC mapping. In this study, we aimed at finding out how two composition methods and spectral–temporal metrics extracted from satellite time series can affect the ability of a machine learning classifier to produce accurate LULC maps. We used the Google Earth Engine (GEE) cloud computing platform to create cloud-free Sentinel-2 (S-2) and Landsat-8 (L-8) time series over the Tehran Province (Iran) as of 2020. Two composition methods, namely, seasonal composites and percentiles metrics, were used to define four datasets based on satellite time series, vegetation indices, and topographic layers. The random forest classifier was used in LULC classification and for identifying the most important variables. Accuracy assessment results showed that the S-2 outperformed the L-8 spectral–temporal metrics at the overall and class level. Moreover, the comparison of composition methods indicated that seasonal composites outperformed percentile metrics in both S-2 and L-8 time series. At the class level, the improved performance of seasonal composites was related to their ability to provide better information about the phenological variation of different LULC classes. Finally, we conclude that this methodology can produce LULC maps based on cloud computing GEE in an accurate and fast way and can be used in large-scale LULC mapping.

Remote Sensing and Spatial Analysis for Land-Take Assessment in Basilicata Region (Southern Italy)

Land use is one of the drivers of land-cover change (LCC) and represents the conversion of natural to artificial land cover. This work aims to describe the land-take-monitoring activities and analyze the development trend in test areas of the Basilicata region. Remote sensing is the primary technique for extracting land-use/land-cover (LULC) data. In this study, a new methodology of classification of Landsat data (TM–OLI) is proposed to detect land-cover information automatically and identify land take to perform a multi-temporal analysis. Moreover, within the defined model, it is crucial to use the territorial information layers of geotopographic database (GTDB) for the detailed definition of the land take. All stages of the classification process were developed using the supervised classification algorithm support vector machine (SVM) change-detection analysis, thus integrating the geographic information system (GIS) remote sensing data and adopting free and open-source software and data. The application of the proposed method allowed us to quickly extract detailed land-take maps with an overall accuracy greater than 90%, reducing the cost and processing time.

Woody Plant Encroachment: Evaluating Methodologies for Semiarid Woody Species Classification from Drone Images

Globally, native semiarid grasslands and savannas have experienced a densification of woody plant species—leading to a multitude of environmental, economic, and cultural changes. These encroached areas are unique in that the diversity of tree species is small, but at the same time the individual species possess diverse phenological responses. The overall goal of this study was to evaluate the ability of very high resolution drone imagery to accurately map species of woody plants encroaching on semiarid grasslands. For a site in the Edwards Plateau ecoregion of central Texas, we used affordable, very high resolution drone imagery to which we applied maximum likelihood (ML), support vector machine (SVM), random forest (RF), and VGG-19 convolutional neural network (CNN) algorithms in combination with pixel-based (with and without post-processing) and object-based (small and large) classification methods. Based on test sample data (n = 1000) the VGG-19 CNN model achieved the highest overall accuracy (96.9%). SVM came in second with an average classification accuracy of 91.2% across all methods, followed by RF (89.7%) and ML (86.8%). Overall, our findings show that RGB drone sensors are indeed capable of providing highly accurate classifications of woody plant species in semiarid landscapes—comparable to and even greater in some regards to those achieved by aerial and drone imagery using hyperspectral sensors in more diverse landscapes.

Modeling of Land Use and Land Cover (LULC) Change Based on Artificial Neural Networks for the Chapecó River Ecological Corridor, Santa Catarina/Brazil

The simulation and analysis of future land use and land cover—LULC scenarios using artificial neural networks (ANN)—has been applied in the last 25 years, producing information for environmental and territorial policy making and implementation. LULC changes have impacts on many levels, e.g., climate change, biodiversity and ecosystem services, soil quality, which, in turn, have implications for the landscape. Therefore, it is fundamental that planning is informed by scientific evidence. The objective of this work was to develop a geographic model to identify the main patterns of LULC transitions between the years 2000 and 2018, to simulate a baseline scenario for the year 2036, and to assess the effectiveness of the Chapecó River ecological corridor (an area created by State Decree No. 2.957/2010), regarding the recovery and conservation of forest remnants and natural fields. The results indicate that the forest remnants have tended to recover their area, systematically replacing silviculture areas. However, natural fields (grassland) are expected to disappear in the near future if proper measures are not taken to protect this ecosystem. If the current agricultural advance pattern is maintained, only 0.5% of natural fields will remain in the ecological corridor by 2036. This LULC trend exposes the low effectiveness of the ecological corridor (EC) in protecting and restoring this vital ecosystem.

Water Information Extraction Based on Multi-Model RF Algorithm and Sentinel-2 Image Data

For the Sentinel-2 multispectral satellite image remote sensing data, due to the rich spatial information, the traditional water body extraction methods cannot meet the needs of practical applications. In this study, a random forest-based RF_16 optimal combination model algorithm is proposed to extract water bodies. The research process uses Sentinel-2 multispectral satellite images and DEM data as the basic data, collected 24 characteristic variable indicators (B2, B3, B4, B8, B11, B12, NDVI, MSAVI, B5, B6, B7, B8A, NDI45, MCARI, REIP, S2REP, IRECI, PSSRa, NDWI, MNDWI, LSWI, DEM, SLOPE, SLOPE ASPECT), and constructed four combined models with different input variables. After analysis, it was determined that RF_16 was the optimal combination for extracting water body information in the study area. Model. The results show that: (1) The characteristic variables that have an important impact on the accuracy of the model are the improved normalized difference water index (MNDWI), band B2 (Blue), normalized water index (NDWI), B4 (Red), B3 (Green), and band B5 (Vegetation Red-Edge 1); (2) The water extraction accuracy of the optimal combined model RF_16 can reach 93.16%, and the Kappa coefficient is 0.8214. The overall accuracy is 0.12% better than the traditional Relief F algorithm. The RF_16 method based on the optimal combination model of random forest is an effective means to obtain high-precision water body information in the study area. It can effectively reduce the “salt and pepper effect” and the influence of mixed pixels such as water and shadows on the water extraction accuracy.

Unmanned Aerial Vehicle (UAV)-Based Remote Sensing for Early-Stage Detection of Ganoderma

Early detection of Basal Stem Rot (BSR) disease in oil palms is an important plantation management activity in Southeast Asia. Practical approaches for the best strategic approach toward the treatment of this disease that originated from Ganoderma Boninense require information about the status of infection. In spite of the availability of conventional methods to detect this disease, they are difficult to be used in plantation areas that are commonly large in terms of planting hectarage; therefore, there is an interest for a quick and delicate technique to facilitate the detection and monitoring of Ganoderma in its early stage. The main goal of this paper is to evaluate the use of remote sensing technique for the rapid detection of Ganoderma-infected oil palms using Unmanned Aerial Vehicle (UAV) imagery integrated with an Artificial Neural Network (ANN) model. Principally, we sought for the most representative mean and standard deviation values from green, red, and near-infrared bands, as well as the best palm circle radius, threshold limit, and the number of hidden neurons for different Ganoderma severity levels. With the obtained modified infrared UAV images at 0.026 m spatial resolution, early BSR infected oil palms were most satisfactorily detected with mean and standard deviation derived from a circle radius of 35 pixels of band green and near-infrared, 1/8 threshold limit, and ANN network by 219 hidden neurons, where the total classification accuracies achieved for training and testing the dataset were 97.52% and 72.73%, respectively. The results from this study signified the utilization of an affordable digital camera and UAV platforms in oil palm plantation, predominantly in disease management. The UAV images integrated with the Levenberg–Marquardt training algorithm illustrated its great potential as an aerial surveillance tool to detect early Ganoderma-infected oil palms in vast plantation areas in a rapid and inexpensive manner.

A Two-Stage Pansharpening Method for the Fusion of Remote-Sensing Images

The pansharpening (PS) of remote-sensing images aims to fuse a high-resolution panchromatic image with several low-resolution multispectral images for obtaining a high-resolution multispectral image. In this work, a two-stage PS model is proposed by integrating the ideas of component replacement and the variational method. The global sparse gradient of the panchromatic image is extracted by variational method, and the weight function is constructed by combining the gradient of multispectral image in which the global sparse gradient can provide more robust gradient information. Furthermore, we refine the results in order to reduce spatial and spectral distortions. Experimental results show that our method had high generalization ability for QuickBird, Gaofen-1, and WorldView-4 satellite data. Experimental results evaluated by seven metrics demonstrate that the proposed two-stage method enhanced spatial details subjective visual effects better than other state-of-the-art methods do. At the same time, in the process of quantitative evaluation, the method in this paper had high improvement compared with that other methods, and some of them can reach a maximal improvement of 60%.

Pan-Sharpening Based on CNN+ Pyramid Transformer by Using No-Reference Loss

The majority of existing deep learning pan-sharpening methods often use simulated degraded reference data due to the missing of real fusion labels which affects the fusion performance. The normally used convolutional neural network (CNN) can only extract the local detail information well which may cause the loss of important global contextual characteristics with long-range dependencies in fusion. To address these issues and to fuse spatial and spectral information with high quality information from the original panchromatic (PAN) and multispectral (MS) images, this paper presents a novel pan-sharpening method by designing the CNN+ pyramid Transformer network with no-reference loss (CPT-noRef). Specifically, the Transformer is used as the main architecture for fusion to supply the global features, the local features in shallow CNN are combined, and the multi-scale features from the pyramid structure adding to the Transformer encoder are learned simultaneously. Our loss function directly learns the spatial information extracted from the PAN image and the spectral information from the MS image which is suitable for the theory of pan-sharpening and makes the network control the spatial and spectral loss simultaneously. Both training and test processes are based on real data, so the simulated degraded reference data is no longer needed, which is quite different from most existing deep learning fusion methods. The proposed CPT-noRef network can effectively solve the huge amount of data required by the Transformer network and extract abundant image features for fusion. In order to assess the effectiveness and universality of the fusion model, we have trained and evaluated the model on the experimental data of WorldView-2(WV-2) and Gaofen-1(GF-1) and compared it with other typical deep learning pan-sharpening methods from both the subjective visual effect and the objective index evaluation. The results show that the proposed CPT-noRef network offers superior performance in both qualitative and quantitative evaluations compared with existing state-of-the-art methods. In addition, our method has the strongest generalization capability by testing the Pleiades and WV-2 images on the network trained by GF-1 data. The no-reference loss function proposed in this paper can greatly enhance the spatial and spectral information of the fusion image with good performance and robustness.

Sentinel-1 Spatiotemporal Simulation Using Convolutional LSTM for Flood Mapping

The synthetic aperture radar (SAR) imagery has been widely applied for flooding mapping based on change detection approaches. However, errors in the mapping result are expected since not all land-cover changes are flood-induced, and those changes are sensitive to SAR data, such as crop growth or harvest over agricultural lands, clearance of forested areas, and/or modifications on the urban landscape. This study, therefore, incorporated historical SAR images to boost the detection of flood-induced changes during extreme weather events, using the Long Short-Term Memory (LSTM) method. Additionally, to incorporate the spatial signatures for the change detection, we applied a deep learning-based spatiotemporal simulation framework, Convolutional Long Short-Term Memory (ConvLSTM), for simulating a synthetic image using Sentinel One intensity time series. This synthetic image will be prepared in advance of flood events, and then it can be used to detect flood areas using change detection when the post-image is available. Practically, significant divergence between the synthetic image and post-image is expected over inundated zones, which can be mapped by applying thresholds to the Delta image (synthetic image minus post-image). We trained and tested our model on three events from Australia, Brazil, and Mozambique. The generated Flood Proxy Maps were compared against reference data derived from Sentinel Two and Planet Labs optical data. To corroborate the effectiveness of the proposed methods, we also generated Delta products for two baseline models (closest post-image minus pre-image and historical mean minus post-image) and two LSTM architectures: normal LSTM and ConvLSTM. Results show that thresholding of ConvLSTM Delta yielded the highest Cohen’s Kappa coefficients in all study cases: 0.92 for Australia, 0.78 for Mozambique, and 0.68 for Brazil. Lower Kappa values obtained in the Mozambique case can be subject to the topographic effect on SAR imagery. These results still confirm the benefits in terms of classification accuracy that convolutional operations provide in time series analysis of satellite data employing spatially correlated information in a deep learning framework.