scholarly journals Potential of Hybrid CNN-RF Model for Early Crop Mapping with Limited Input Data

2021 ◽  
Vol 13 (9) ◽  
pp. 1629
Author(s):  
Geun-Ho Kwak ◽  
Chan-won Park ◽  
Kyung-do Lee ◽  
Sang-il Na ◽  
Ho-yong Ahn ◽  
...  
Keyword(s):  
Time Series ◽  
Input Data ◽  
Training Data ◽  
Sufficient Time ◽  
Early Crop ◽  
Crop Mapping ◽  
Crop Classification ◽  
Limited Input Data ◽  
Time Series Images ◽  
And Training

When sufficient time-series images and training data are unavailable for crop classification, features extracted from convolutional neural network (CNN)-based representative learning may not provide useful information to discriminate crops with similar spectral characteristics, leading to poor classification accuracy. In particular, limited input data are the main obstacles to obtain reliable classification results for early crop mapping. This study investigates the potential of a hybrid classification approach, i.e., CNN-random forest (CNN-RF), in the context of early crop mapping, that combines the automatic feature extraction capability of CNN with the superior discrimination capability of an RF classifier. Two experiments on incremental crop classification with unmanned aerial vehicle images were conducted to compare the performance of CNN-RF with that of CNN and RF with respect to the length of the time-series and training data sizes. When sufficient time-series images and training data were used for the classification, the accuracy of CNN-RF was slightly higher or comparable with that of CNN. In contrast, when fewer images and the smallest training data were used at the early crop growth stage, CNN-RF was substantially beneficial and the overall accuracy increased by maximum 6.7%p and 4.6%p in the two study areas, respectively, compared to CNN. This is attributed to its ability to discriminate crops from features with insufficient information using a more sophisticated classifier. The experimental results demonstrate that CNN-RF is an effective classifier for early crop mapping when only limited input images and training samples are available.

scholarly journals Forest-Type Classification Using Time-Weighted Dynamic Time Warping Analysis in Mountain Areas: A Case Study in Southern China

Forests ◽  
2019 ◽  
Vol 10 (11) ◽  
pp. 1040 ◽  
Author(s):  
Kai Cheng ◽  
Juanle Wang
Keyword(s):  
Time Series ◽  
Dynamic Time Warping ◽  
Forest Type ◽  
Southern China ◽  
Training Data ◽  
Mountain Forest ◽  
Time Warping ◽  
Mountain Areas ◽  
Dynamic Time ◽  
Time Series Images

Efficient methodologies for mapping forest types in complicated mountain areas are essential for the implementation of sustainable forest management practices and monitoring. Existing solutions dedicated to forest-type mapping are primarily focused on supervised machine learning algorithms (MLAs) using remote sensing time-series images. However, MLAs are challenged by complex and problematic forest type compositions, lack of training data, loss of temporal data caused by clouds obscuration, and selection of input feature sets for mountainous areas. The time-weighted dynamic time warping (TWDTW) is a supervised classifier, an adaptation of the dynamic time warping method for time series analysis for land cover classification. This study evaluates the performance of the TWDTW method that uses a combination of Sentinel-2 and Landsat-8 time-series images when applied to complicated mountain forest-type classifications in southern China with complex topographic conditions and forest-type compositions. The classification outputs were compared to those produced by MLAs, including random forest (RF) and support vector machine (SVM). The results presented that the three forest-type maps obtained by TWDTW, RF, and SVM have high consistency in spatial distribution. TWDTW outperformed SVM and RF with mean overall accuracy and mean kappa coefficient of 93.81% and 0.93, respectively, followed by RF and SVM. Compared with MLAs, TWDTW method achieved the higher classification accuracy than RF and SVM, with even less training data. This proved the robustness and less sensitivities to training samples of the TWDTW method when applied to mountain forest-type classifications.

scholarly journals Feature Selection of Time Series MODIS Data for Early Crop Classification Using Random Forest: A Case Study in Kansas, USA

2015 ◽  
Vol 7 (5) ◽  
pp. 5347-5369 ◽  
Author(s):  
Pengyu Hao ◽  
Yulin Zhan ◽  
Li Wang ◽  
Zheng Niu ◽  
Muhammad Shakir
Keyword(s):  
Time Series ◽  
Feature Selection ◽  
Random Forest ◽  
Modis Data ◽  
Early Crop ◽  
Crop Classification ◽  
Selection Of

An Adaptive Method to Further Improve the Tolerance of an ANN-Based Detection System for Mooring Line Failure

2021 ◽  
Author(s):  
Djoni E. Sidarta ◽  
Nicolas Tcherniguin ◽  
Ho-Joon Lim ◽  
Philippe Bouchard ◽  
Mengchen Kang ◽  
...  
Keyword(s):  
Time Series ◽  
Adaptive Method ◽  
Training Data ◽  
Computational Time ◽  
Mooring Line ◽  
Ann Model ◽  
Hydrodynamic Simulations ◽  
Sensitivity Tests ◽  
Fully Coupled ◽  
And Training

Abstract The use of an Artificial Neural Network (ANN) for detection of mooring line failure has been a growing subject of discussion over the past several years. Sidarta et al. [6, 8, 12] have presented papers on the detection of mooring line failure of a moored vessel by monitoring shifts in the low frequency periods, mean yaw angles as a function of vessel positions, mass and added mass. An ANN model has been trained using MLTSIM hydrodynamic simulations based on information from the early stages of the project. The restoring forces and moments from mooring lines, risers and umbilicals have been solved using catenary equations to significantly reduce the computational time to generate the ANN training data. This paper presents the evaluation of this ANN model using fully coupled OrcaFlex hydrodynamic simulations, based on the latest information of the project. The results of this evaluation demonstrate the tolerance of the trained ANN model as it can properly function when tested using time series of vessel motions from the fully coupled OrcaFlex hydrodynamic simulations. Furthermore, although the ANN model has been trained using simulations with a completely removed line, the trained model can still function when tested with simulations of a line broken at the bottom. These give affirmation that the ANN model can tolerate the differences that exist between the test and training data. Sensitivity of the polyester line stiffness has also been performed using fully coupled OrcaFlex hydrodynamic simulations, and the computed time series of vessel motions have been used to test the ANN model. The ANN model can deal with some level of differences between the sensitivity tests and training data. However, sensitivity tests of the polyester line stiffness to model aging lines has posed a real challenge to the ANN model as its prediction accuracy has decreased significantly. This paper presents an adaptive method that can be implemented such that the ANN model can adapt to relatively new conditions that are quite different from the training data and maintain the accuracy of its prediction. With this approach, an existing ANN model that has been trained under certain assumptions of the system can still function although the behavior of the system has drifted away from those assumptions. This phenomenon may have similarity with a possible reality that measured behavior in the field can be somewhat different from numerical simulations. This adaptive method has a potential for addressing this issue such that a simulation trained ANN model can maintain its expected accuracy although dealing with different conditions from the training data. If successful, this is a good cost saving scenario that an ANN model adapts to some degree to relatively new and different conditions before the differences become too much to handle and the only solution is to retrain the model.

scholarly journals Evaluation of Three Deep Learning Models for Early Crop Classification Using Sentinel-1A Imagery Time Series—A Case Study in Zhanjiang, China

2019 ◽  
Vol 11 (22) ◽  
pp. 2673 ◽  
Author(s):  
Hongwei Zhao ◽  
Zhongxin Chen ◽  
Hao Jiang ◽  
Wenlong Jing ◽  
Liang Sun ◽  
...  
Keyword(s):  
Time Series ◽  
Deep Learning ◽  
Optical Remote Sensing ◽  
Learning Models ◽  
Learning Methods ◽  
Early Crop ◽  
Crop Classification ◽  
Incremental Classification ◽  
Time Point

Timely and accurate estimation of the area and distribution of crops is vital for food security. Optical remote sensing has been a key technique for acquiring crop area and conditions on regional to global scales, but great challenges arise due to frequent cloudy days in southern China. This makes optical remote sensing images usually unavailable. Synthetic aperture radar (SAR) could bridge this gap since it is less affected by clouds. The recent availability of Sentinel-1A (S1A) SAR imagery with a 12-day revisit period at a high spatial resolution of about 10 m makes it possible to fully utilize phenological information to improve early crop classification. In deep learning methods, one-dimensional convolutional neural networks (1D CNNs), long short-term memory recurrent neural networks (LSTM RNNs), and gated recurrent unit RNNs (GRU RNNs) have been shown to efficiently extract temporal features for classification tasks. However, due to the complexity of training, these three deep learning methods have been less used in early crop classification. In this work, we attempted to combine them with an incremental classification method to avoid the need for training optimal architectures and hyper-parameters for data from each time series. First, we trained 1D CNNs, LSTM RNNs, and GRU RNNs based on the full images’ time series to attain three classifiers with optimal architectures and hyper-parameters. Then, starting at the first time point, we performed an incremental classification process to train each classifier using all of the previous data, and obtained a classification network with all parameter values (including the hyper-parameters) at each time point. Finally, test accuracies of each time point were assessed for each crop type to determine the optimal time series length. A case study was conducted in Suixi and Leizhou counties of Zhanjiang City, China. To verify the effectiveness of this method, we also implemented the classic random forest (RF) approach. The results were as follows: (i) 1D CNNs achieved the highest Kappa coefficient (0.942) of the four classifiers, and the highest value (0.934) in the GRU RNNs time series was attained earlier than with other classifiers; (ii) all three deep learning methods and the RF achieved F measures above 0.900 before the end of growth seasons of banana, eucalyptus, second-season paddy rice, and sugarcane; while, the 1D CNN classifier was the only one that could obtain an F-measure above 0.900 for pineapple before harvest. All results indicated the effectiveness of the solution combining the deep learning models with the incremental classification approach for early crop classification. This method is expected to provide new perspectives for early mapping of croplands in cloudy areas.

Cloning Safe Driving Behavior for Self-Driving Cars using Convolutional Neural Networks

2019 ◽  
Vol 12 (2) ◽  
pp. 120-127 ◽  
Author(s):  
Wael Farag
Keyword(s):  
Gradient Descent ◽  
Autonomous Driving ◽  
Driving Behavior ◽  
Training Data ◽  
Stochastic Gradient Descent ◽  
Data Set ◽  
Safe Driving ◽  
Processing Pipeline ◽  
Self Driving Cars ◽  
And Training

Background: In this paper, a Convolutional Neural Network (CNN) to learn safe driving behavior and smooth steering manoeuvring, is proposed as an empowerment of autonomous driving technologies. The training data is collected from a front-facing camera and the steering commands issued by an experienced driver driving in traffic as well as urban roads. Methods: This data is then used to train the proposed CNN to facilitate what it is called “Behavioral Cloning”. The proposed Behavior Cloning CNN is named as “BCNet”, and its deep seventeen-layer architecture has been selected after extensive trials. The BCNet got trained using Adam’s optimization algorithm as a variant of the Stochastic Gradient Descent (SGD) technique. Results: The paper goes through the development and training process in details and shows the image processing pipeline harnessed in the development. Conclusion: The proposed approach proved successful in cloning the driving behavior embedded in the training data set after extensive simulations.

scholarly journals Bernoulli Time Series Modelling with Application to Accommodation Tourism Demand

2021 ◽  
Vol 5 (1) ◽  
pp. 17
Author(s):  
Miguel Ángel Ruiz Reina
Keyword(s):  
Time Series ◽  
Time Series Data ◽  
Prognostic Models ◽  
Series Data ◽  
Economic Agents ◽  
Time Series Modelling ◽  
Simulation And Training ◽  
Uncertainty Method ◽  
And Training ◽  
Hotel Accommodation

In this research, a new uncertainty method has been developed and applied to forecasting the hotel accommodation market. The simulation and training of Time Series data are from January 2001 to December 2018 in the Spanish case. The Log-log BeTSUF method estimated by GMM-HAC-Newey-West is considered as a contribution for measuring uncertainty vs. other prognostic models in the literature. The results of our model present better indicators of the RMSE and Ratio Theil’s for the predictive evaluation period of twelve months. Furthermore, the straightforward interpretation of the model and the high descriptive capacity of the model allow economic agents to make efficient decisions.

scholarly journals The Performance of Post-Fall Detection Using the Cross-Dataset: Feature Vectors, Classifiers and Processing Conditions

Sensors ◽  
2021 ◽  
Vol 21 (14) ◽  
pp. 4638
Author(s):  
Bummo Koo ◽  
Jongman Kim ◽  
Yejin Nam ◽  
Youngho Kim
Keyword(s):  
Time Series ◽  
Fall Detection ◽  
Discrete Data ◽  
Classification Performance ◽  
Training Data ◽  
Processing Conditions ◽  
Feature Vectors ◽  
External Data ◽  
Public Dataset ◽  
The Cross

In this study, algorithms to detect post-falls were evaluated using the cross-dataset according to feature vectors (time-series and discrete data), classifiers (ANN and SVM), and four different processing conditions (normalization, equalization, increase in the number of training data, and additional training with external data). Three-axis acceleration and angular velocity data were obtained from 30 healthy male subjects by attaching an IMU to the middle of the left and right anterior superior iliac spines (ASIS). Internal and external tests were performed using our lab dataset and SisFall public dataset, respectively. The results showed that ANN and SVM were suitable for the time-series and discrete data, respectively. The classification performance generally decreased, and thus, specific feature vectors from the raw data were necessary when untrained motions were tested using a public dataset. Normalization made SVM and ANN more and less effective, respectively. Equalization increased the sensitivity, even though it did not improve the overall performance. The increase in the number of training data also improved the classification performance. Machine learning was vulnerable to untrained motions, and data of various movements were needed for the training.

scholarly journals Simulation of Daily Mean Soil Temperatures for Agricultural Land Use Considering Limited Input Data

Atmosphere ◽  
2021 ◽  
Vol 12 (4) ◽  
pp. 441
Author(s):  
Philipp Grabenweger ◽  
Branislava Lalic ◽  
Miroslav Trnka ◽  
Jan Balek ◽  
Erwin Murer ◽  
...  
Keyword(s):  
Soil Temperature ◽  
Input Data ◽  
Agricultural Land ◽  
Soil Column ◽  
Soil Conditions ◽  
Snow Layer ◽  
Time Step ◽  
Soil Temperatures ◽  
Dimensional Simulation ◽  
Limited Input Data

A one-dimensional simulation model that simulates daily mean soil temperature on a daily time-step basis, named AGRISOTES (AGRIcultural SOil TEmperature Simulation), is described. It considers ground coverage by biomass or a snow layer and accounts for the freeze/thaw effect of soil water. The model is designed for use on agricultural land with limited (and mostly easily available) input data, for estimating soil temperature spatial patterns, for single sites (as a stand-alone version), or in context with agrometeorological and agronomic models. The calibration and validation of the model are carried out on measured soil temperatures in experimental fields and other measurement sites with various climates, agricultural land uses and soil conditions in Europe. The model validation shows good results, but they are determined strongly by the quality and representativeness of the measured or estimated input parameters to which the model is most sensitive, particularly soil cover dynamics (biomass and snow cover), soil pore volume, soil texture and water content over the soil column.

scholarly journals Robust Approach to Supervised Deep Neural Network Training for Real-Time Object Classification in Cluttered Indoor Environment

2021 ◽  
Vol 11 (15) ◽  
pp. 7148
Author(s):  
Bedada Endale ◽  
Abera Tullu ◽  
Hayoung Shi ◽  
Beom-Soo Kang
Keyword(s):  
Neural Network ◽  
Deep Learning ◽  
Real Time ◽  
Network Architecture ◽  
Input Data ◽  
Deep Neural Network ◽  
Data Augmentation ◽  
Object Classification ◽  
Training Data ◽  
Gradient Descent Algorithm

Unmanned aerial vehicles (UAVs) are being widely utilized for various missions: in both civilian and military sectors. Many of these missions demand UAVs to acquire artificial intelligence about the environments they are navigating in. This perception can be realized by training a computing machine to classify objects in the environment. One of the well known machine training approaches is supervised deep learning, which enables a machine to classify objects. However, supervised deep learning comes with huge sacrifice in terms of time and computational resources. Collecting big input data, pre-training processes, such as labeling training data, and the need for a high performance computer for training are some of the challenges that supervised deep learning poses. To address these setbacks, this study proposes mission specific input data augmentation techniques and the design of light-weight deep neural network architecture that is capable of real-time object classification. Semi-direct visual odometry (SVO) data of augmented images are used to train the network for object classification. Ten classes of 10,000 different images in each class were used as input data where 80% were for training the network and the remaining 20% were used for network validation. For the optimization of the designed deep neural network, a sequential gradient descent algorithm was implemented. This algorithm has the advantage of handling redundancy in the data more efficiently than other algorithms.

Sign in / Sign up

Export Citation Format

Share Document