scholarly journals Mapping Rice Paddy Based on Machine Learning with Sentinel-2 Multi-Temporal Data: Model Comparison and Transferability

2020 ◽  
Vol 12 (10) ◽  
pp. 1620 ◽  
Author(s):  
Weichun Zhang ◽  
Hongbin Liu ◽  
Wei Wu ◽  
Linqing Zhan ◽  
Jing Wei
Keyword(s):  
Machine Learning ◽  
Remote Sensing ◽  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Rice Paddy ◽  
Temporal Data ◽  
Model Transferability ◽  
Local Classification ◽  
Multi Temporal ◽  
Sentinel 2

Rice is an important agricultural crop in the Southwest Hilly Area, China, but there has been a lack of efficient and accurate monitoring methods in the region. Recently, convolutional neural networks (CNNs) have obtained considerable achievements in the remote sensing community. However, it has not been widely used in mapping a rice paddy, and most studies lack the comparison of classification effectiveness and efficiency between CNNs and other classic machine learning models and their transferability. This study aims to develop various machine learning classification models with remote sensing data for comparing the local accuracy of classifiers and evaluating the transferability of pretrained classifiers. Therefore, two types of experiments were designed: local classification experiments and model transferability experiments. These experiments were conducted using cloud-free Sentinel-2 multi-temporal data in Banan District and Zhongxian County, typical hilly areas of Southwestern China. A pure pixel extraction algorithm was designed based on land-use vector data and a Google Earth Online image. Four convolutional neural network (CNN) algorithms (one-dimensional (Conv-1D), two-dimensional (Conv-2D) and three-dimensional (Conv-3D_1 and Conv-3D_2) convolutional neural networks) were developed and compared with four widely used classifiers (random forest (RF), extreme gradient boosting (XGBoost), support vector machine (SVM) and multilayer perceptron (MLP)). Recall, precision, overall accuracy (OA) and F1 score were applied to evaluate classification accuracy. The results showed that Conv-2D performed best in local classification experiments with OA of 93.14% and F1 score of 0.8552 in Banan District, OA of 92.53% and F1 score of 0.8399 in Zhongxian County. CNN-based models except Conv-1D provided more desirable performance than non-CNN classifiers. Besides, among the non-CNN classifiers, XGBoost received the best result with OA of 89.73% and F1 score of 0.7742 in Banan District, SVM received the best result with OA of 88.57% and F1 score of 0.7538 in Zhongxian County. In model transferability experiments, almost all CNN classifiers had low transferability. RF and XGBoost models have achieved acceptable F1 scores for transfer (RF = 0.6673 and 0.6469, XGBoost = 0.7171 and 0.6709, respectively).

scholarly journals 3D Convolutional Neural Networks for Crop Classification with Multi-Temporal Remote Sensing Images

2018 ◽  
Vol 10 (2) ◽  
pp. 75 ◽  
Author(s):  
Shunping Ji ◽  
Chi Zhang ◽  
Anjian Xu ◽  
Yun Shi ◽  
Yulin Duan
Keyword(s):  
Remote Sensing ◽  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Remote Sensing Images ◽  
Multi Temporal ◽  
Crop Classification

scholarly journals LITHOLOGICAL CLASSIFICATION USING MULTI-SENSOR DATA AND CONVOLUTIONAL NEURAL NETWORKS

2019 ◽  
Vol XLII-2/W16 ◽  
pp. 55-59 ◽  
Author(s):  
M. Brandmeier ◽  
Y. Chen
Keyword(s):  
Remote Sensing ◽  
Neural Networks ◽  
Deep Learning ◽  
Convolutional Neural Networks ◽  
Sensor Data ◽  
Support Vector ◽  
Carbonate Sediments ◽  
Feature Maps ◽  
Aster Data ◽  
Sentinel 2

<p><strong>Abstract.</strong> Deep learning has been used successfully in computer vision problems, e.g. image classification, target detection and many more. We use deep learning in conjunction with ArcGIS to implement a model with advanced convolutional neural networks (CNN) for lithological mapping in the Mount Isa region (Australia). The area is ideal for spectral remote sensing as there is only sparse vegetation and besides freely available Sentinel-2 and ASTER data, several geophysical datasets are available from exploration campaigns. By fusing the data and thus covering a wide spectral range as well as capturing geophysical properties of rocks, we aim at improving classification accuracies and support geological mapping. We also evaluate the performance of the sensors on their own compared to a joint use as the Sentinel-2 satellites are relatively new and as of now there exist only few studies for geological applications. We developed an end-to-end deep learning model using Keras and Tensorflow that consists of several convolutional, pooling and deconvolutional layers. Our model was inspired by the family of U-Net architectures, where low-level feature maps (encoders) are concatenated with high-level ones (decoders), which enables precise localization. This type of network architecture was especially designed to effectively solve pixel-wise classification problems, which is appropriate for lithological classification. We spatially resampled and fused the multi-sensor remote sensing data with different bands and geophysical data into image cubes as input for our model. Pre-processing was done in ArcGIS and the final, fine-tuned model was imported into a toolbox to be used on further scenes directly in the GIS environment. The tool classifies each pixel of the multiband imagery into different types of rocks according to a defined probability threshold. Results highlight the power of using Sentinel-2 in conjunction with ASTER data with accuracies of 75% in comparison to only 70% and 73% for ASTER or Sentinel-2 data alone. These results are similar but examining the different classes shows that there are significant improvements for classes such as dolerite or carbonate sediments that are not that widely distributed in the area. Adding geophysical datasets reduced accuracies to 60%, probably due to an order of magnitude difference in spatial resolution. In comparison, Random Forest (RF) and Support Vector Machines (SVMs) that were trained on the same data only achieve accuracies of 46 % and 36 % respectively. Most insecurity is due to labelling errors and labels with mixed lithologies. However, results show that the U-Netmodel is a powerful alternative to other classifiers for medium-resolution multispectral data.</p>

scholarly journals Assessing the Effect of Training Sampling Design on the Performance of Machine Learning Classifiers for Land Cover Mapping Using Multi-Temporal Remote Sensing Data and Google Earth Engine

2021 ◽  
Vol 13 (8) ◽  
pp. 1433
Author(s):  
Shobitha Shetty ◽  
Prasun Kumar Gupta ◽  
Mariana Belgiu ◽  
S. K. Srivastav
Keyword(s):  
Machine Learning ◽  
Remote Sensing ◽  
Random Sampling ◽  
Sampling Design ◽  
Remote Sensing Data ◽  
Google Earth ◽  
Machine Learning Classifiers ◽  
Learning Classifiers ◽  
Multi Temporal ◽  
Google Earth Engine

Machine learning classifiers are being increasingly used nowadays for Land Use and Land Cover (LULC) mapping from remote sensing images. However, arriving at the right choice of classifier requires understanding the main factors influencing their performance. The present study investigated firstly the effect of training sampling design on the classification results obtained by Random Forest (RF) classifier and, secondly, it compared its performance with other machine learning classifiers for LULC mapping using multi-temporal satellite remote sensing data and the Google Earth Engine (GEE) platform. We evaluated the impact of three sampling methods, namely Stratified Equal Random Sampling (SRS(Eq)), Stratified Proportional Random Sampling (SRS(Prop)), and Stratified Systematic Sampling (SSS) upon the classification results obtained by the RF trained LULC model. Our results showed that the SRS(Prop) method favors major classes while achieving good overall accuracy. The SRS(Eq) method provides good class-level accuracies, even for minority classes, whereas the SSS method performs well for areas with large intra-class variability. Toward evaluating the performance of machine learning classifiers, RF outperformed Classification and Regression Trees (CART), Support Vector Machine (SVM), and Relevance Vector Machine (RVM) with a >95% confidence level. The performance of CART and SVM classifiers were found to be similar. RVM achieved good classification results with a limited number of training samples.

Sign in / Sign up

Export Citation Format

Share Document