Zhiwen Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Zhiwen Li has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Xiaoxuan Ma
School of Electrical and Information Engineering, Beijing University of Civil Engineering and Architecture, Beijing, China
2 shared publications - Hengyou Wang
School of Science, Beijing University of Civil Engineering and Architecture, Beijing, China
1 shared publication - Lei Zhang
School of Intelligent Science and Technology, Beijing University of Civil Engineering and Architecture, Beijing 100044, China
1 shared publication Yibo Deng
School of Electrical and Information Engineering, Beijing University of Civil Engineering and Architecture, China
1 shared publication
Latest Publications
- Open Access
- 3 April 2023
- Published by Agora University of Oradea in International Journal of Computers Communications & Control
- Vol. 18 (2)
A Novel Generative Image Inpainting Model with Dense Gated Convolutional Network
Damaged image inpainting is one of the hottest research fields in computer image processing. The development of deep learning, especially Convolutional Neural Network (CNN), has significantly enhanced the effect of image inpainting. However, the direct connection between convolution layers may increase the risk of gradient disappearance or overfitting during training process. In addition, pixel artifacts or visual inconsistencies may occur if the damaged area is inpainted directly. To solve the above problems, we propose a novel Dense Gated Convolutional Network (DGCN) for generative image inpainting by modifying the gated convolutional network structure in this paper. Firstly, Holistically-nested edge detector (HED) is utilized to predict the edge information of the missing areas to assist the subsequent inpainting task to reduce the generation of artifacts. Then, dense connections are added to the generative network to reduce the network parameters while reducing the risk of instability in the training process. Finally, the experimental results on CelebA and Places2 datasets show that the proposed model achieves better inpainting results in terms of PSNR, SSIM and visual effects compared with other classical image inpainting models. DGCN has the common advantages of gated convolution and dense connection, which can reduce network parameters and improve the inpainting effect of the network.
- 3 April 2023
- Published by Agora University of Oradea in International Journal of Computers Communications & Control
- Vol. 18 (2)
Open AccessA Novel Generative Image Inpainting Model with Dense Gated Convolutional Network
Damaged image inpainting is one of the hottest research fields in computer image processing. The development of deep learning, especially Convolutional Neural Network (CNN), has significantly enhanced the effect of image inpainting. However, the direct connection between convolution layers may increase the risk of gradient disappearance or overfitting during training process. In addition, pixel artifacts or visual inconsistencies may occur if the damaged area is inpainted directly. To solve the above problems, we propose a novel Dense Gated Convolutional Network (DGCN) for generative image inpainting by modifying the gated convolutional network structure in this paper. Firstly, Holistically-nested edge detector (HED) is utilized to predict the edge information of the missing areas to assist the subsequent inpainting task to reduce the generation of artifacts. Then, dense connections are added to the generative network to reduce the network parameters while reducing the risk of instability in the training process. Finally, the experimental results on CelebA and Places2 datasets show that the proposed model achieves better inpainting results in terms of PSNR, SSIM and visual effects compared with other classical image inpainting models. DGCN has the common advantages of gated convolution and dense connection, which can reduce network parameters and improve the inpainting effect of the network.
- Open Access
Deep Matrix Factorization Based on Convolutional Neural Networks for Image Inpainting
In this work, we formulate the image in-painting as a matrix completion problem. Traditional matrix completion methods are generally based on linear models, assuming that the matrix is low rank. When the original matrix is large scale and the observed elements are few, they will easily lead to over-fitting and their performance will also decrease significantly. Recently, researchers have tried to apply deep learning and nonlinear techniques to solve matrix completion. However, most of the existing deep learning-based methods restore each column or row of the matrix independently, which loses the global structure information of the matrix and therefore does not achieve the expected results in the image in-painting. In this paper, we propose a deep matrix factorization completion network (DMFCNet) for image in-painting by combining deep learning and a traditional matrix completion model. The main idea of DMFCNet is to map iterative updates of variables from a traditional matrix completion model into a fixed depth neural network. The potential relationships between observed matrix data are learned in a trainable end-to-end manner, which leads to a high-performance and easy-to-deploy nonlinear solution. Experimental results show that DMFCNet can provide higher matrix completion accuracy than the state-of-the-art matrix completion methods in a shorter running time.
Open AccessDeep Matrix Factorization Based on Convolutional Neural Networks for Image Inpainting
In this work, we formulate the image in-painting as a matrix completion problem. Traditional matrix completion methods are generally based on linear models, assuming that the matrix is low rank. When the original matrix is large scale and the observed elements are few, they will easily lead to over-fitting and their performance will also decrease significantly. Recently, researchers have tried to apply deep learning and nonlinear techniques to solve matrix completion. However, most of the existing deep learning-based methods restore each column or row of the matrix independently, which loses the global structure information of the matrix and therefore does not achieve the expected results in the image in-painting. In this paper, we propose a deep matrix factorization completion network (DMFCNet) for image in-painting by combining deep learning and a traditional matrix completion model. The main idea of DMFCNet is to map iterative updates of variables from a traditional matrix completion model into a fixed depth neural network. The potential relationships between observed matrix data are learned in a trainable end-to-end manner, which leads to a high-performance and easy-to-deploy nonlinear solution. Experimental results show that DMFCNet can provide higher matrix completion accuracy than the state-of-the-art matrix completion methods in a shorter running time.