Transferring Pre-Trained Deep CNNs for Remote Scene Classification with General Features Learned from Linear PCA Network

Wang, Jie; Luo, Chang; Huang, Hanqiao; Zhao, Haitao; Wang, Shiqiang

doi:10.3390/rs9030225

Cited by 56 publications

(25 citation statements)

References 54 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…It can be also noted that the proposed method outperforms other deep models, such as GoogLeNet [7] which obtains 97.10%, VGG-VD16-1 st -FC+Aug [23] which obtains 96.88%, SPP-net+MKL [29] which obtains 96.38%, and MCNN [21] which obtains 96.66%. For Brazilian Coffee Scene dataset, the proposed method also obtains 91.24% outperforms 88.46% which is obtained by LQPCANet [37], 85.36% by VGG16 [22], 89.79% by ConvNet [38], 90.94% by CaffeNet [7] and 91.13% by D-DSML-CaffeNet [9]. For Google dataset, the proposed method obtains 92.04% which is better than 82.81% by TF-CNN [39], 89.88% by RDSG-CNN [39], 87.68% by Fine-tuned CaffeNet.…”

Section: Comparisons With the Most Recent Methodsmentioning

confidence: 78%

Joint Learning of the Center Points and Deep Metrics for Land-Use Classification in Remote Sensing

Gong

Zhong

et al. 2019

Remote Sensing

View full text Add to dashboard Cite

Deep learning methods, especially convolutional neural networks (CNNs), have shown remarkable ability for remote sensing scene classification. However, the traditional training process of standard CNNs only takes the point-wise penalization of the training samples into consideration, which usually makes the learned CNNs sub-optimal especially for remote sensing scenes with large intra-class variance and low inter-class variance. To address this problem, deep metric learning, which incorporates the metric learning into the deep model, is used to maximize the inter-class variance and minimize the intra-class variance for better representation. This work introduces structured metric learning for remote sensing scene representation, a special deep metric learning which can take full advantage of the training batch. However, the deep metrics only consider the pairwise correlation between the training samples, and ignores the classwise correlation from the class view. To take the classwise penalization into consideration, this work defines the center points of the learned features of each class in the training process to represent the class. Through increasing the variance between different center points and decreasing the variance between the learned features from each class and the corresponding center point, the representational ability can be further improved. Therefore, this work develops a novel center-based structured metric learning to take advantage of both the deep metrics and the center points. Finally, joint supervision of the cross-entropy loss and the center-based structured metric learning is developed for the land-use classification in remote sensing. It can joint learn the center points and the deep metrics to take advantage of the point-wise, the pairwise, and the classwise correlation. Experiments are conducted over three real-world remote sensing scene datasets, namely UC Merced Land-Use dataset, Brazilian Coffee Scene dataset, and Google dataset. The classification performance can achieve 97.30%, 91.24%, and 92.04% with the proposed method over the three datasets which are better than other state-of-the-art methods under the same experimental setups. The results demonstrate that the proposed method can improve the representational ability for the remote sensing scenes.

show abstract

Section: Comparisons With the Most Recent Methodsmentioning

confidence: 78%

Joint Learning of the Center Points and Deep Metrics for Land-Use Classification in Remote Sensing

Gong

Zhong

et al. 2019

Remote Sensing

View full text Add to dashboard Cite

show abstract

“…Mask-RCNN requires a large amount of annotated data for training to avoid overfitting. To overcome the problem of limited annotated dataset in remote sensing domain, we adopted transfer learning by selected the pre-trained network weights of the resnet50 model, which was successfully trained with the image net dataset [36]. We utilized the pre-trained resnet50 and fine-tuned the network weights to the NWPUVHR dataset.…”

Section: Training Phasementioning

confidence: 99%

Object Detection Using Adaptive Mask RCNN in Optical Remote Sensing Images

Mahmoud¹,

Mohamed²,

El-Khoribi³

et al. 2020

IJIES

View full text Add to dashboard Cite

Fast and automatic object detection in remote sensing images is a critical and challenging task for civilian and military applications. Recently, deep learning approaches were introduced to overcome the limitation of traditional object detection methods. In this paper, adaptive mask Region-based Convolutional Network (mask-RCNN) is utilized for multi-class object detection in remote sensing images. Transfer learning, data augmentation, and fine-tuning were adopted to overcome objects scale variability, small size, the density of objects, and the scarcity of annotated remote sensing image. Also, five optimization methods were investigated namely: Adaptive Moment Estimation (Adam), stochastic gradient decent (SGD), adaptive learning rate method (Adelta), Root Mean Square Propagation (RMSprop) and hybrid optimization. In hybrid optimization, the training process begins Adam then switches to SGD when appropriate and vice versa. Also, the behaviour of adaptive mask RCNN was compared to baseline deep object detection methods. Several experiments were conducted on the challenging NWPU-VHR-10 dataset. The hybrid method Adam_SGD acheived the highest Accuracy precision, with 95%. Experimental results showed detection performance in terms of accuracy and intersection over union (IOU) boost of performance up to 6%.

show abstract

“…The experimental performance showed that the proposed algorithm was superior to the classical Reed-Xiaoli [26] and the most advanced representation-based detectors, such as sparse representation-based detectors (SRD) [27] and cooperative representation-based detectors [28]. Wang proposed two architectures to extract the general features of remote scene classification from the pre-trained CNNs [29]. Wang Liwei et al proposed a hyperspectral image classification method by applying transfer learning in deep residual networks [30], and shared the shallow network weight parameters of deep residual networks.…”

Section: Introductionmentioning

confidence: 99%

Deep Transfer HSI Classification Method Based on Information Measure and Optimal Neighborhood Noise Reduction

et al. 2019

View full text Add to dashboard Cite

Land environment is one of the most commonly and importantly used synthetical natural environments in a virtual test. To recognize the ground truth for the construction of virtual land environment, a deep transfer hyperspectral image (HSI) classification method based on information measure and optimal neighborhood noise reduction was proposed in this article. Firstly, the information measure method was used to select the most valuable spectrum. Specifically, three representative bands were selected using the combination of entropy, color matching function, and mutual information. Based on the selected bands, a patch containing spatial-spectral information was constructed and used as the input of the convolutional neural networks (CNN) network. Then, in order to address the problem that a large number of labeled samples were required in deep learning method, the HSI classification method based on deep transfer learning was proposed. In the proposed method, the transfer of parameters ensured the classification performance with small training samples and reduced the training cost. Moreover, the optimal neighborhood noise reduction was used as the post-processing method to effectively eliminate the salt-and-pepper noise and further improve the classification performance. Experiments on two datasets demonstrated that the proposed method in this article had higher classification accuracy than similar methods.

show abstract

Transferring Pre-Trained Deep CNNs for Remote Scene Classification with General Features Learned from Linear PCA Network

Cited by 56 publications

References 54 publications

Joint Learning of the Center Points and Deep Metrics for Land-Use Classification in Remote Sensing

Joint Learning of the Center Points and Deep Metrics for Land-Use Classification in Remote Sensing

Object Detection Using Adaptive Mask RCNN in Optical Remote Sensing Images

Deep Transfer HSI Classification Method Based on Information Measure and Optimal Neighborhood Noise Reduction

Contact Info

Product

Resources

About