An Optimized Faster R-CNN Method Based on DRNet and RoI Align for Building Detection in Remote Sensing Images

Bai, Tong; Pang, Yu; Wang, Junchao; Han, Kaining; Luo, Jiasai; Hui-qian, Wang; Lin, Jie; Wu, Jun; Zhang, Hui

doi:10.3390/rs12050762

Cited by 53 publications

(34 citation statements)

References 27 publications

(26 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…At present, ANN has already developed into various new structures. The main structure of this study is the Convolutional Neural Network (CNN) [24, 25], and three other neural network structures are compared with it. These four neural network structures are detailed in the following subsections.…”

Section: Artificial Neural Network: a Backgroundmentioning

confidence: 99%

Multiple-Input Deep Convolutional Neural Network Model for COVID-19 Forecasting in China

Huang

Chen

et al. 2020

Preprint

100

View full text Add to dashboard Cite

COVID-19 is spreading all across the globe. Up until March 23, 2020, the confirmed cases in 173 12 countries and regions of the globe had surpassed 346,000, and more than 14,700 deaths had resulted. The 13 confirmed cases outside of China had also reached over 81,000, with over 3,200 deaths. In this study, a 14 Convolutional Neural Network (CNN) was proposed to analyze and predict the number of confirmed cases. 15 Several cities with the most confirmed cases in China were the focus of this study, and a COVID-19 16 forecasting model, based on the CNN deep neural network method, was proposed. To compare the overall 17 efficacies of different algorithms, the indicators of mean absolute error and root mean square error were 18 applied in the experiment of this study. The experiment results indicated that compared with other deep 19 learning methods, the CNN model proposed in this study has the greatest prediction efficacy. The feasibility 20 and practicality of the model in predicting the cumulative number of COVID-19 confirmed cases were also 21 verified in this study.22

show abstract

Section: Artificial Neural Network: a Backgroundmentioning

confidence: 99%

Multiple-Input Deep Convolutional Neural Network Model for COVID-19 Forecasting in China

Huang

Chen

et al. 2020

Preprint

100

View full text Add to dashboard Cite

show abstract

“…As a result of their excellent classification and regression capabilities, CNNs [39][40][41] are widely applied in image processing. Basically, they can extract features from original data using digital filters, before using fully connected layers for classification or regression-based prediction.…”

Section: Convolutional Neural Network (Cnn)mentioning

confidence: 99%

A novel hybrid deep neural network model for short‐term electricity price forecasting

et al. 2020

View full text Add to dashboard Cite

Summary A Ubiquitous Power Internet of Things is fundamentally an Internet of Things, but focused upon power systems. Being able to predict these prices accurately may help with the identification of customer needs and the effective regulation of the power grid by power producers. It may also help electric power traders to manage risks, make correct decisions, and obtain more benefits. In this paper, a novel hybrid model is proposed for short‐term electricity price prediction. The model consists of three algorithms: Variational Mode Decomposition (VMD); a Convolutional Neural Network (CNN); and Gated Recurrent Unit (GRU). This is called SEPNet for convenience. The annual electricity price data is divided into seasons because of seasonal differences in the time series of electricity prices. The VMD algorithm is used to decompose the complex time series of electricity prices into intrinsic mode functions (IMFs) with different center frequencies. The CNN is used to further extract the time‐domain features for all the intrinsic model functions in the VMD domain. The GRU is then employed to process and learn the time‐domain features extracted by the CNN, leading to the final prediction. A comparison is made with five models, such as LSTM, CNN, VMD‐CNN, BP, VMD‐ELMAN. The results showed that the proposed model had the best performance, and it was found that using VMD can improve the Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE) for the four seasons by 84% and 81%, respectively. The addition of GRU in the SEPNet model further improved the MAPE and RMSE by 19% and 25%, respectively. Including CNN and VMD‐CNN, that shows that the proposed model has the best performance. The MAPE and RMSE for the four seasonal averages are 0.730% and 0.453, respectively. This confirms that the SEPNet model has the feasibility and high accuracy to predict short‐term electricity prices.

show abstract

“…Algorithms have been used to detect roof tile buildings, flat building detection has been used to detect non-tile flat buildings according to shape features, and results fusion has been used to fuse and aggregate results from previous blocks [30]. The Tong Bai team uses the improved algorithm Faster R-CNN (region-based Convolutional Neural Network), which adopts DRNet (Dense Residual Network) and RoI (Region of Interest) Align to utilize texture information and to solve the region mismatch problems [31]. Another example of a solution to this problem is the work of the Evangelos Maltezos team, which used LiDAR [Light Detection and Ranging] data to detect buildings [32].…”

Section: Introductionmentioning

confidence: 99%

Detection, Classification and Boundary Regularization of Buildings in Satellite Imagery Using Faster Edge Region Convolutional Neural Networks

Reda¹,

Kędzierski²

2020

Remote Sensing

View full text Add to dashboard Cite

With the development of effective deep learning algorithms, it became possible to achieve high accuracy when conducting remote sensing analyses on very high-resolution images (VHRS), especially in the context of building detection and classification. In this article, in order to improve the accuracy of building detection and classification, we propose a Faster Edge Region Convolutional Neural Networks (FER-CNN) algorithm. This proposed algorithm is trained and evaluated on different datasets. In addition, we propose a new method to improve the detection of the boundaries of detected buildings. The results of our algorithm are compared with those of other methods, such as classical Faster Region Convolution Neural Network (Faster R-CNN) with the original VGG16 and the Single-Shot Multibox Detector (SSD). The experimental results show that our methods make it possible to obtain an average detection accuracy of 97.5% with a false positive classification rate of 8.4%. An additional advantage of our method is better resistance to shadows, which is a very common issue for satellite images of urban areas. Future research will include designing and training the neural network to detect small buildings, as well as irregularly shaped buildings that are partially obscured by shadows or other occlusions.

show abstract

An Optimized Faster R-CNN Method Based on DRNet and RoI Align for Building Detection in Remote Sensing Images

Cited by 53 publications

References 27 publications

Multiple-Input Deep Convolutional Neural Network Model for COVID-19 Forecasting in China

Multiple-Input Deep Convolutional Neural Network Model for COVID-19 Forecasting in China

A novel hybrid deep neural network model for short‐term electricity price forecasting

Detection, Classification and Boundary Regularization of Buildings in Satellite Imagery Using Faster Edge Region Convolutional Neural Networks

Contact Info

Product

Resources

About