Yitong Yan

Ms. Yitong Yan

orcid

Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
12 March 2021 in IEEE Transactions on Multimedia
Verified
Journal Article
Closed access
YYYitong Yan
Yitong Yan
CLChuangchuang Liu
Chuangchuang Liu
CCChangyou Chen
Changyou Chen
XSXianfang Sun
Xianfang Sun
LJLongcun Jin
Longcun Jin
XZXiang Zhou
Xiang Zhou
Traditional super-resolution (SR) methods by minimize the mean square error usually produce images with over-smoothed and blurry edges, due to the lack of high-frequency details. In this paper, we propose two novel techniques within the generative adversarial network framework to encourage generation of photo-realistic images for image super-resolution. Firstly, instead of producing a single score to discriminate real and fake images, we propose a variant, called Fine-grained Attention Generative Adversarial Network (FASRGAN), to discriminate each pixel of real and fake images. FASRGAN adopts a UNet-like network as the discriminator with two outputs: an image score and an image score map. The score map has the same spatial size as the HR/SR images, serving as the fine-grained attention to represent the degree of reconstruction difficulty for each pixel. Secondly, instead of using different networks for the generator and the discriminator, we introduce a feature-sharing variant (denoted as Fs-SRGAN) for both the generator and the discriminator. The sharing mechanism can maintain model express power while making the model more compact, and thus can improve the ability of producing high-quality images. Quantitative and visual comparisons with state-of-the-art methods on benchmark datasets demonstrate the superiority of our methods. We further apply our super-resolution images for object recognition, which further demonstrates the effectiveness of our proposed method. The code is available at https://github.com/Rainyfish/FASRGAN-and-Fs-SRGAN .

30 March 2021 in IEEE Access
Verified
Journal Article
Open access
YYYitong Yan
Yitong Yan
XXXue Xu
Xue Xu
WCWenhui Chen
Wenhui Chen
XPXinyi Peng
Xinyi Peng
Recently, deep convolutional neural networks (CNN) have been widely applied in the single image super-resolution (SISR) task and achieved significant progress in reconstruction performance. However, most of the existing CNN-based SR models are impractical to real-world applicants due to numerous parameters and heavy computation. To tackle this issue, we propose a lightweight attended multi-scale residual network (LAMRN) in this work. Specially, we present an attended multi-scale residual block (AMSRB) to extract multi-scale features, where we embed the efficient channel attention block (ECA) to enhance the discrimination of features. Besides, we introduce a double-attention fusion (DAF) block to fuse the low-level and high-level features efficiently. We use spatial attention and channel attention to obtain guidance from the low-level and high-level features, which is used to guide the feature fusion. Extensive experimental results demonstrate that our LAMRN achieves competitive performance against the state-of-the-art methods with similar parameters and computational operations.

2 March 2020 in IEEE Access
Verified
Journal Article
Open access
JQJiayv Qin
Jiayv Qin
XSXianfang Sun
Xianfang Sun
YYYitong Yan
Yitong Yan
LJLongcun Jin
Longcun Jin
XPXinyi Peng
Xinyi Peng
With the help of deep convolutional neural networks, a vast majority of single image super-resolution (SISR) methods have been developed, and achieved promising performance. However, these methods suffer from over-smoothness in textured regions due to utilizing a single-resolution network to reconstruct both the low-frequency and high-frequency information simultaneously. To overcome this problem, we propose a Multi-resolution space-Attended Residual Dense Network (MARDN) to separate low-frequency and high-frequency information for reconstructing high-quality super-resolved images. Specifically, we start from a low-resolution sub-network, and add low-to-high resolution sub-networks step by step in several stages. These sub-networks with different depth and resolution are utilized to produce feature maps of different frequencies in parallel. For instance, the high-resolution sub-network with fewer stages is applied to local high-frequency textured information extraction, while the low-resolution one with more stages is devoted to generating global low-frequency information. Furthermore, the fusion block with channel-wise sub-network attention is proposed for adaptively fusing the feature maps from different sub-networks instead of applying concatenation and 1×1 convolution. A series of ablation investigations and model analyses validate the effectiveness and efficiency of our MARDN. Extensive experiments on benchmark datasets demonstrate the superiority of the proposed MARDN against the state-of-the-art methods. Our super-resolution results and the source code can be downloaded from https://github.com/Periter/MARDN.

20 May 2020 in IEEE Access
Verified
Journal Article
Open access
WCWenhui Chen
Wenhui Chen
CLChuangchuang Liu
Chuangchuang Liu
YYYitong Yan
Yitong Yan
LJLongcun Jin
Longcun Jin
XSXianfang Sun
Xianfang Sun
XPXinyi Peng
Xinyi Peng
The PSNR-oriented super-resolution (SR) methods pursue high reconstruction accuracy, but tend to produce over-smoothed results and lose plenty of high-frequency details. The GAN-based SR methods aim to generate more photo-realistic images, but the hallucinatory details are often accompanied with unsatisfying artifacts and noise. To address these problems, we propose a guided dual super-resolution network (GDSR), which exploits the advantages of both the PSNR-oriented and the GAN-based methods to achieve a good trade-off between reconstruction accuracy and perceptual quality. Specifically, our network contains two branches, where one is trained to extract global information and the other to focus on detail information. In this way, our network simultaneously generates SR images with high accuracy and satisfactory visual quality. To obtain more high-frequency features, we use the global features extracted from the low-frequency branch to guide the training of the high-frequency branch. Besides, our method utilizes a mask network to adaptively recover the final super-resolved image. Extensive experiments on several standard benchmarks show that our proposed method achieves better performance compared with state-of-the-art methods. The source code and the results of our GDSR are available at https://github.com/wenchen4321/GDSR.