scholarly journals High-Resolution U-Net: Preserving Image Details for Cultivated Land Extraction

Sensors ◽  
2020 ◽  
Vol 20 (15) ◽  
pp. 4064
Author(s):  
Wenna Xu ◽  
Xinping Deng ◽  
Shanxin Guo ◽  
Jinsong Chen ◽  
Luyi Sun ◽  
...  
Keyword(s):  
High Resolution ◽  
Random Forest ◽  
Land Cover ◽  
Image Texture ◽  
Support Vector ◽  
Spectral Variation ◽  
Cultivated Land ◽  
Resource Monitoring ◽  
K Nearest Neighbors ◽  
Image Details

Accurate and efficient extraction of cultivated land data is of great significance for agricultural resource monitoring and national food security. Deep-learning-based classification of remote-sensing images overcomes the two difficulties of traditional learning methods (e.g., support vector machine (SVM), K-nearest neighbors (KNN), and random forest (RF)) when extracting the cultivated land: (1) the limited performance when extracting the same land-cover type with the high intra-class spectral variation, such as cultivated land with both vegetation and non-vegetation cover, and (2) the limited generalization ability for handling a large dataset to apply the model to different locations. However, the “pooling” process in most deep convolutional networks, which attempts to enlarge the sensing field of the kernel by involving the upscale process, leads to significant detail loss in the output, including the edges, gradients, and image texture details. To solve this problem, in this study we proposed a new end-to-end extraction algorithm, a high-resolution U-Net (HRU-Net), to preserve the image details by improving the skip connection structure and the loss function of the original U-Net. The proposed HRU-Net was tested in Xinjiang Province, China to extract the cultivated land from Landsat Thematic Mapper (TM) images. The result showed that the HRU-Net achieved better performance (Acc: 92.81%; kappa: 0.81; F1-score: 0.90) than the U-Net++ (Acc: 91.74%; kappa: 0.79; F1-score: 0.89), the original U-Net (Acc: 89.83%; kappa: 0.74; F1-score: 0.86), and the Random Forest model (Acc: 76.13%; kappa: 0.48; F1-score: 0.69). The robustness of the proposed model for the intra-class spectral variation and the accuracy of the edge details were also compared, and this showed that the HRU-Net obtained more accurate edge details and had less influence from the intra-class spectral variation. The model proposed in this study can be further applied to other land cover types that have more spectral diversity and require more details of extraction.

scholarly journals High-resolution digital mapping of soil organic carbon in permafrost terrain using machine learning: a case study in a sub-Arctic peatland environment

2018 ◽  
Vol 15 (6) ◽  
pp. 1663-1682 ◽  
Author(s):  
Matthias B. Siewert
Keyword(s):  
Machine Learning ◽  
High Resolution ◽  
Random Forest ◽  
Organic Carbon ◽  
Soil Organic Carbon ◽  
Land Cover ◽  
Support Vector ◽  
Small Scale ◽  
Permafrost Degradation ◽  
Discontinuous Permafrost

Abstract. Soil organic carbon (SOC) stored in northern peatlands and permafrost-affected soils are key components in the global carbon cycle. This article quantifies SOC stocks in a sub-Arctic mountainous peatland environment in the discontinuous permafrost zone in Abisko, northern Sweden. Four machine-learning techniques are evaluated for SOC quantification: multiple linear regression, artificial neural networks, support vector machine and random forest. The random forest model performed best and was used to predict SOC for several depth increments at a spatial resolution of 1 m (1×1 m). A high-resolution (1 m) land cover classification generated for this study is the most relevant predictive variable. The landscape mean SOC storage (0–150 cm) is estimated to be 8.3 ± 8.0 kg C m−2 and the SOC stored in the top meter (0–100 cm) to be 7.7 ± 6.2 kg C m−2. The predictive modeling highlights the relative importance of wetland areas and in particular peat plateaus for the landscape's SOC storage. The total SOC was also predicted at reduced spatial resolutions of 2, 10, 30, 100, 250 and 1000 m and shows a significant drop in land cover class detail and a tendency to underestimate the SOC at resolutions  >  30 m. This is associated with the occurrence of many small-scale wetlands forming local hot-spots of SOC storage that are omitted at coarse resolutions. Sharp transitions in SOC storage associated with land cover and permafrost distribution are the most challenging methodological aspect. However, in this study, at local, regional and circum-Arctic scales, the main factor limiting robust SOC mapping efforts is the scarcity of soil pedon data from across the entire environmental space. For the Abisko region, past SOC and permafrost dynamics indicate that most of the SOC is barely 2000 years old and very dynamic. Future research needs to investigate the geomorphic response of permafrost degradation and the fate of SOC across all landscape compartments in post-permafrost landscapes.

scholarly journals High-resolution digital mapping of soil organic carbon in permafrost terrain using machine-learning: A case study in a sub-Arctic peatland environment

2017 ◽  
Author(s):  
Matthias B. Siewert
Keyword(s):  
Machine Learning ◽  
High Resolution ◽  
Random Forest ◽  
Organic Carbon ◽  
Soil Organic Carbon ◽  
Land Cover ◽  
Support Vector ◽  
Small Scale ◽  
Permafrost Degradation ◽  
Discontinuous Permafrost

Abstract. Soil organic carbon (SOC) stored in northern peatlands and permafrost affected soils are key components in the global carbon cycle. I quantify SOC stocks in a sub-arctic mountainous peatland environment in the discontinuous permafrost zone in Abisko, northern Sweden. Four machine-learning techniques are evaluated: multiple linear regression, artificial neural networks, support vector machine and random forest. The random forest approach performed best and was used to predict SOC for several depth increments at a spatial resolution of 2 ×2 m. A high-resolution (1 × 1 m) land cover classification generated for this study is the most relevant predictive variable. The landscape mean SOC storage (0–150 cm) is estimated to 7.9 ± 8.0 kg C m−2 and the SOC stored in the top meter (0–100 cm) to 7.0 ± 6.3 kg C m−2. The predictive modeling highlights the relative importance of wetland areas and in particular peat plateaus for the landscape SOC storage. A surprising large number of small scale wetland areas are mapped forming very local hot-spots of SOC storage. The results show that robust SOC predictions are possible with the available methods and very high-resolution remote sensing data. Strong environmental gradients associated with land cover and permafrost distribution are the most challenging methodological aspect. However, in this study, at local, regional and circum-Arctic scale the main factor limiting robust, high-resolution SOC mapping efforts is the scarcity of soil pedon data from across the entire environmental space. For the Absiko region, past SOC and permafrost dynamics indicate that most of the SOC is barely 2000 years old and very dynamic in wetland areas with permafrost related landforms. Future research needs to investigate the geomorphic response of permafrost degradation and the fate of SOC across all landscape compartments in post-permafrost landscapes.

scholarly journals Classifications of Breast Cancer Diagnosis using Machine Learning

2020 ◽  
Vol 14 ◽  
Keyword(s):  
Breast Cancer ◽  
Machine Learning ◽  
Random Forest ◽  
Breast Cancer Diagnosis ◽  
Performance Comparison ◽  
Support Vector ◽  
Breast Cancer Dataset ◽  
K Nearest Neighbors ◽  
Cancer Dataset ◽  
Machine Learning Classification

Breast Cancer (BC) is amongst the most common and leading causes of deaths in women throughout the world. Recently, classification and data analysis tools are being widely used in the medical field for diagnosis, prognosis and decision making to help lower down the risks of people dying or suffering from diseases. Advanced machine learning methods have proven to give hope for patients as this has helped the doctors in early detection of diseases like Breast Cancer that can be fatal, in support with providing accurate outcomes. However, the results highly depend on the techniques used for feature selection and classification which will produce a strong machine learning model. In this paper, a performance comparison is conducted using four classifiers which are Multilayer Perceptron (MLP), Support Vector Machine (SVM), K-Nearest Neighbors (KNN) and Random Forest on the Wisconsin Breast Cancer dataset to spot the most effective predictors. The main goal is to apply best machine learning classification methods to predict the Breast Cancer as benign or malignant using terms such as accuracy, f-measure, precision and recall. Experimental results show that Random forest is proven to achieve the highest accuracy of 99.26% on this dataset and features, while SVM and KNN show 97.78% and 97.04% accuracy respectively. MLP shows the least accuracy of 94.07%. All the experiments are conducted using RStudio as the data mining tool platform.

scholarly journals Deep learning approach for classifying, detecting and predicting photometric redshifts of quasars in the Sloan Digital Sky Survey stripe 82

2018 ◽  
Vol 611 ◽  
pp. A97 ◽  
Author(s):  
J. Pasquet-Itam ◽  
J. Pasquet
Keyword(s):  
Random Forest ◽  
Random Forest Classifier ◽  
Sloan Digital Sky Survey ◽  
Light Curves ◽  
Support Vector ◽  
Learning Approach ◽  
Photometric Redshifts ◽  
K Nearest Neighbors ◽  
Sky Survey ◽  
Extraction Step

We have applied a convolutional neural network (CNN) to classify and detect quasars in the Sloan Digital Sky Survey Stripe 82 and also to predict the photometric redshifts of quasars. The network takes the variability of objects into account by converting light curves into images. The width of the images, noted w, corresponds to the five magnitudes ugriz and the height of the images, noted h, represents the date of the observation. The CNN provides good results since its precision is 0.988 for a recall of 0.90, compared to a precision of 0.985 for the same recall with a random forest classifier. Moreover 175 new quasar candidates are found with the CNN considering a fixed recall of 0.97. The combination of probabilities given by the CNN and the random forest makes good performance even better with a precision of 0.99 for a recall of 0.90. For the redshift predictions, the CNN presents excellent results which are higher than those obtained with a feature extraction step and different classifiers (a K-nearest-neighbors, a support vector machine, a random forest and a Gaussian process classifier). Indeed, the accuracy of the CNN within |Δz| < 0.1 can reach 78.09%, within |Δz| < 0.2 reaches 86.15%, within |Δz| < 0.3 reaches 91.2% and the value of root mean square (rms) is 0.359. The performance of the KNN decreases for the three |Δz| regions, since within the accuracy of |Δz| < 0.1, |Δz| < 0.2, and |Δz| < 0.3 is 73.72%, 82.46%, and 90.09% respectively, and the value of rms amounts to 0.395. So the CNN successfully reduces the dispersion and the catastrophic redshifts of quasars. This new method is very promising for the future of big databases such as the Large Synoptic Survey Telescope.

scholarly journals Glass Classification based on Machine Learning Algorithms

2020 ◽  
Vol 9 (11) ◽  
pp. 139-142
Keyword(s):  
Machine Learning ◽  
Random Forest ◽  
Amorphous Solid ◽  
Machine Learning Algorithms ◽  
Support Vector ◽  
K Nearest Neighbors ◽  
X Ray ◽  
Svm Algorithm ◽  
Artificial Neural Network Ann ◽  
Logistic Regression Algorithm

Glass Industry is considered one of the most important industries in the world. The Glass is used everywhere, from water bottles to X-Ray and Gamma Rays protection. This is a non-crystalline, amorphous solid that is most often transparent. There are lots of uses of glass, and during investigation in a crime scene, the investigators need to know what is type of glass in a scene. To find out the type of glass, we will use the online dataset and machine learning to solve the above problem. We will be using ML algorithms such as Artificial Neural Network (ANN), K-nearest neighbors (KNN) algorithm, Support Vector Machine (SVM) algorithm, Random Forest algorithm, and Logistic Regression algorithm. By comparing all the algorithm Random Forest did the best in glass classification.

scholarly journals High-Resolution Mining-Induced Geo-Hazard Mapping Using Random Forest: A Case Study of Liaojiaping Orefield, Central China

2021 ◽  
Vol 13 (18) ◽  
pp. 3638
Author(s):  
Yaozu Qin ◽  
Li Cao ◽  
Ali Darvishi Boloorani ◽  
Weicheng Wu
Keyword(s):  
High Resolution ◽  
Random Forest ◽  
Land Cover ◽  
Human Life ◽  
Ensemble Classifier ◽  
Central China ◽  
Sensor Data ◽  
Environmental Drivers ◽  
Hazard Mapping ◽  
High Susceptibility

Mining-induced geo-hazard mapping (MGM) is a critical step for reducing and avoiding tremendous losses of human life, mine production, and property that are caused by ore mining. Due to the restriction of the survey techniques and data sources, high-resolution MGM remains a big challenge. To overcome this problem, in this research, such an MGM was conducted using detailed geological exploration and topographic survey data as well as Gaofen-1 satellite imagery as multi-source geoscience datasets and machine learning technique taking Liaojiaping Orefield, Central China as an example. First, using Gaofen-1 panchromatic and multispectral (PMS) sensor data and Random Forest (RF) non-parametric ensemble classifier, a seven-class land cover map was generated for the study area with an overall accuracy (OA) and Kappa coefficient (KC) of 99.69% and 98.37%, respectively. Next, several environmental drivers including land cover, topography (aspect and slope), lithology, distance from fault, elevation difference between surface and underground excavation, and the difference of spectral information from PMS multispectral data of different years were integrated as predictors to construct an RF-based MGM model. The constructed model showed an excellent prediction performance, with an OA of 98.53%, KC of 97.06%, and AUC of 0.998, and the 85.60% of the observed geo-disaster that have occurred in the predicted high susceptibility class (encompassing 2.82% of the study area). The results suggested that the changes in environmental factors in the high susceptibility areas can be used as indicators for monitoring and early-warning of the geo-disaster occurrence.

scholarly journals A Novel Class-Specific Object-Based Method for Urban Change Detection Using High-Resolution Remote Sensing Imagery

2021 ◽  
Vol 87 (4) ◽  
pp. 249-262
Author(s):  
Ting Bai ◽  
Kaimin Sun ◽  
Wenzhuo Li ◽  
Deren Li ◽  
Yepei Chen ◽  
...  
Keyword(s):  
Support Vector Machine ◽  
High Resolution ◽  
Land Cover ◽  
Change Detection ◽  
Urban Areas ◽  
Support Vector ◽  
Change Type ◽  
Optimal Classifier ◽  
Object Based ◽  
Specific Object

A single-scale object-based change-detection classifier can distinguish only global changes in land cover, not the more granular and local changes in urban areas. To overcome this issue, a novel class-specific object-based change-detection method is proposed. This method includes three steps: class-specific scale selection, class-specific classifier selection, and land cover change detection. The first step combines multi-resolution segmentation and a random forest to select the optimal scale for each change type in land cover. The second step links multi-scale hierarchical sampling with a classifier such as random forest, support vector machine, gradient-boosting decision tree, or Adaboost; the algorithm automatically selects the optimal classifier for each change type in land cover. The final step employs the optimal classifier to detect binary changes and from-to changes for each change type in land cover. To validate the proposed method, we applied it to two high-resolution data sets in urban areas and compared the change-detection results of our proposed method with that of principal component analysis k-means, object-based change vector analysis, and support vector machine. The experimental results show that our proposed method is more accurate than the other methods. The proposed method can address the high levels of complexity found in urban areas, although it requires historical land cover maps as auxiliary data.

Sign in / Sign up

Export Citation Format

Share Document