The image annotation algorithm using convolutional features from intermediate layer of deep learning

Chen, Yuantao; Liu, Linwu; Tao, Jiajun; Chen, Xi; Xia, Runlong; Zhang, Qian; Xiong, Jie; Yang, Kai; Xie, Jiang

doi:10.1007/s11042-020-09887-2

Cited by 64 publications

(23 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the minimization process, a proper optimization strategy is employed to find the optimal network parameters W that minimize the objective function (7) to obtain the required latent HR image. Because the objective function may not be fully differentiable, it can be divided into smooth and nonsmooth parts through the proximal gradient algorithm [26].…”

Section: Problem Formulation and Minimization Strategymentioning

confidence: 99%

RBDN: Residual Bottleneck Dense Network for Image Super-Resolution

Zhang

Sheng

et al. 2021

IEEE Access

View full text Add to dashboard Cite

Recent studies have shown that a super-resolution generative adversarial network (SRGAN) can significantly improve the quality of single-image super-resolution. However, existing SRGAN methods also have certain drawbacks, such as an insufficient feature utilization, a large number of parameters. To further enhance the visual quality, we thoroughly studied three key components of SRGAN, i.e., the network architecture, adversarial loss, and perceptual loss, and propose a DenseNet with Residual-in-Residual Bottleneck Block (RRBB), called a residual bottleneck dense network (RBDN), for single-image superresolution. First, to improve the utilization of features between the various layers of the network, we adopted a dense cascading connection between layers. At the same time, to reduce the computational cost, we added a bottleneck structure to each layer, greatly reducing the number of network parameters and accelerating the convergence speed of the training process. Second, the proposed RRBB, as the basic network building unit, removes the batch normalization (BN) layer and employs the ELU function to reduce the opposite effects in the absence of BN. In addition, we applied an improved overall loss function during the model training process to stably train the model and further improve the realism of the reconstructed high-resolution image. To prove the superiority of our proposed model, we conducted a comprehensive and objective evaluation of the Peak Signal-to-Noise Ratio, structural similarity, learned perceptual image patch similarity, and other evaluation indicators obtained from the three test sets, i.e., Set5, Set14, and BSD100, from the recent stateof-the-art model. Finally, we conducted qualitative and quantitative analyses of the results obtained in terms of the evaluation indicators, the authenticity of the restored HR images, and textural details, which show the superiority of the RBDN model.

show abstract

Section: Problem Formulation and Minimization Strategymentioning

confidence: 99%

RBDN: Residual Bottleneck Dense Network for Image Super-Resolution

Zhang

Sheng

et al. 2021

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Super resolution plays an important role in the field of artificial intelligence [1][2]. This paper focuses on the introduction of Single Image Super-Resolution (SISR) technology, which has been widely used in image compression, medical imaging [3][4][5], remote sensing imaging [6], public security [7] and other fields due to its flexibility, simplicity and high practicability. It is a research hotspot in the field of image processing.…”

Section: Introductionmentioning

confidence: 99%

A Super Resolution Algorithm Based on Attention Mechanism and SRGAN Network

Liu¹,

Chen²

2021

IEEE Access

View full text Add to dashboard Cite

Image super-resolution reconstruction uses a specific algorithm to restore the low resolution blurred image in the same scene to a high resolution image. In recent years, with the vigorous development of deep learning, this technology has been widely used in many fields. In the field of image super-resolution reconstruction, more and more methods based on deep learning have been studied. According to the principle of GAN, a pseudo high-resolution image is generated by the generator, and then the discriminator calculates the difference between the image and the real high-resolution image to measure the authenticity of the image. Based on SRGAN (super resolution general adverse network), this paper mainly makes three improvements. First, it introduces the attention channel mechanism, that is, it adds Ca (channel attention) module to SRGAN network, and increases the network depth to better express high frequency features; Second, delete the original BN (batch normalization) layer to improve the network performance; Third, modify the loss function to reduce the impact of noise on the image. The experimental results show that the proposed method is superior to the current methods in both quantitative and qualitative indicators, and promotes the recovery of high-frequency detail information. The experimental results show that the proposed method improves the artifact problem and improves the PSNR (peak signal-to-noise ratio) on set5, set10 and bsd100 test sets.

show abstract

“…In recent years, the application of deep neural networks has spread across many fields, its powerful feature extraction and feature representation capabilities have enabled it to achieve impressive results in various fields. For example, in the field of verification code recognition, Wang et al [9] used the DenseNet model and adopted cross-layer connections to improve the recognition accuracy while reducing the problem of gradient disappearance and reducing the number of parameters; Chen et al [10] based on the deep learning method, through the intermediate layer of the pretrained deep learning model to output the convolution results, combined with the positive mean vector method to establish a visual feature vector database, to achieve automatic image annotation. At the same time, image hiding based on deep neural networks has also appeared in recent years.…”

Section: Introductionmentioning

confidence: 99%

StegoPNet: Image Steganography With Generalization Ability Based on Pyramid Pooling Module

et al. 2020

View full text Add to dashboard Cite

In terms of payload capacity and visual effects, the existing image steganography technology based on deep neural networks still needs improvement, to solve this problem, this paper proposes a new deep convolutional steganography network based on the pyramid pooling module to achieve better image steganography. The deep convolutional neural network itself can extract features efficiently. Based on the combination of up-sampling structure, we added a pyramid pool module, under the premise of ensuring safety, fully integrated the previous important global features, achieved good hiding and extraction effects, fully integrated the previous important global features, and effective it reduces the loss of contextual information between different sub-regions in the feature extraction process and achieves better hiding and extraction effects under the premise of ensuring security. Experiments show that the average peak signalto-noise ratio (PSNR)/structure similarity (SSIM) and other indicators between the images obtained by this method have achieved good results in the experiment. Also, we have verified through ablation experiments that the pyramid pooling module can enhance the steganography effect of the network model and can further cut down the loss function of the model. INDEX TERMS Image steganography, deep neural network, pyramid pooling module.

show abstract

The image annotation algorithm using convolutional features from intermediate layer of deep learning

Cited by 64 publications

References 34 publications

RBDN: Residual Bottleneck Dense Network for Image Super-Resolution

RBDN: Residual Bottleneck Dense Network for Image Super-Resolution

A Super Resolution Algorithm Based on Attention Mechanism and SRGAN Network

StegoPNet: Image Steganography With Generalization Ability Based on Pyramid Pooling Module

Contact Info

Product

Resources

About