Pub Date : 2022-08-01DOI: 10.1109/cost57098.2022.00034
Yaxi Wang, Ziyun Tong, Xin Chen, Yuyang Wang
There are many problems with the age-appropriate design of the existing application (APP) interface. The author improves the design of the Railroad 12306 Love Edition APP, and proposes an improved design and output on the basis of the original. Using theoretical comparative analysis and usability analysis, the usability problems of Railroad 12306 Love Edition APP are derived, the priority of user needs is derived through card classification method, and the usability evaluation of the output design results is finally conducted. Based on the original design, the design results were output through analysis methods, and the final design was optimized for the station big screen and online ticket booking of the Railroad 12306 Love APP. This approach improves the efficiency of use by older adults, reduces error rates, and improves subjective satisfaction with APP use. It provides new reference cases and reference ideas for other APP interfaces to adapt to aging transformation.
{"title":"Usability Study of Railroad 12306 Love Edition Application Based on User Experience","authors":"Yaxi Wang, Ziyun Tong, Xin Chen, Yuyang Wang","doi":"10.1109/cost57098.2022.00034","DOIUrl":"https://doi.org/10.1109/cost57098.2022.00034","url":null,"abstract":"There are many problems with the age-appropriate design of the existing application (APP) interface. The author improves the design of the Railroad 12306 Love Edition APP, and proposes an improved design and output on the basis of the original. Using theoretical comparative analysis and usability analysis, the usability problems of Railroad 12306 Love Edition APP are derived, the priority of user needs is derived through card classification method, and the usability evaluation of the output design results is finally conducted. Based on the original design, the design results were output through analysis methods, and the final design was optimized for the station big screen and online ticket booking of the Railroad 12306 Love APP. This approach improves the efficiency of use by older adults, reduces error rates, and improves subjective satisfaction with APP use. It provides new reference cases and reference ideas for other APP interfaces to adapt to aging transformation.","PeriodicalId":135595,"journal":{"name":"2022 International Conference on Culture-Oriented Science and Technology (CoST)","volume":"52 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128193226","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-01DOI: 10.1109/cost57098.2022.00053
Qian Chen, Siving Du, Haonan Yang, Xiao Gong
Most of the current literature research on electronic user consumption behavior believes that game payment behavior mainly comes from “digital identity” in the game, and this identity agree mainly depends on the time spent by users in games. The more time users spend building this digital identity, the more willing they are to pay for games. In game design, game companies will also focus on how to promote game users to stay online longer in the game through the improvement of game quality. After the rise of eSports, electronic games are no longer the only medium between companies and users. DOTA2 “interactive compendium” is a new medium used in games to communicate with users. The interactive compendium exists on the basis of The International DOTA2 Championships(TI), which has a paid impact on users who do not play games. This study takes the DOTA2 “cloud player” group as the research object, and analyzes the main factors affecting their consumption through surveys, especially factors outside the game, explore how the “digital identity” of users is preserved and continues to function. Which proves that users may continue to pay for games even if they no longer use game products, or even pay more, but a new medium is required.
{"title":"Analysis on Digital Identity and Consumption Behavior of DOTA2 “Cloud Players”","authors":"Qian Chen, Siving Du, Haonan Yang, Xiao Gong","doi":"10.1109/cost57098.2022.00053","DOIUrl":"https://doi.org/10.1109/cost57098.2022.00053","url":null,"abstract":"Most of the current literature research on electronic user consumption behavior believes that game payment behavior mainly comes from “digital identity” in the game, and this identity agree mainly depends on the time spent by users in games. The more time users spend building this digital identity, the more willing they are to pay for games. In game design, game companies will also focus on how to promote game users to stay online longer in the game through the improvement of game quality. After the rise of eSports, electronic games are no longer the only medium between companies and users. DOTA2 “interactive compendium” is a new medium used in games to communicate with users. The interactive compendium exists on the basis of The International DOTA2 Championships(TI), which has a paid impact on users who do not play games. This study takes the DOTA2 “cloud player” group as the research object, and analyzes the main factors affecting their consumption through surveys, especially factors outside the game, explore how the “digital identity” of users is preserved and continues to function. Which proves that users may continue to pay for games even if they no longer use game products, or even pay more, but a new medium is required.","PeriodicalId":135595,"journal":{"name":"2022 International Conference on Culture-Oriented Science and Technology (CoST)","volume":"58 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127048009","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-01DOI: 10.1109/cost57098.2022.00019
Yijie Zhou, Gang Wu, Qiang Lin, Dingguo Yu, Hui Wu
With the prevailing of deep learning technology, automatic virtual image synthesis has made huge progress and the popularity of virtual portraits has been growing rapidly. Traditional virtual synthesis system rely on computer graphics method driven by motion capture of a real person, which need labor and equipment costs. In view of this, our paper proposes a virtual host synthesis method based on text driven to generate lip shape and facial animation(include eye movement and head pose) from a signal facial image of a virtual host. More precisely, we use three main modules to construct a virtual host synthesis system: a speech synthesis module based on Tacotron2, a speech to landmark points module to extract mixture landmarks movement to speech, and video generation module based on conditional generative adversarial network to generate video frames and realize time-continuous automatic sport news reporting.
{"title":"Text-based Talking Facial Synthesis for Virtual Host System","authors":"Yijie Zhou, Gang Wu, Qiang Lin, Dingguo Yu, Hui Wu","doi":"10.1109/cost57098.2022.00019","DOIUrl":"https://doi.org/10.1109/cost57098.2022.00019","url":null,"abstract":"With the prevailing of deep learning technology, automatic virtual image synthesis has made huge progress and the popularity of virtual portraits has been growing rapidly. Traditional virtual synthesis system rely on computer graphics method driven by motion capture of a real person, which need labor and equipment costs. In view of this, our paper proposes a virtual host synthesis method based on text driven to generate lip shape and facial animation(include eye movement and head pose) from a signal facial image of a virtual host. More precisely, we use three main modules to construct a virtual host synthesis system: a speech synthesis module based on Tacotron2, a speech to landmark points module to extract mixture landmarks movement to speech, and video generation module based on conditional generative adversarial network to generate video frames and realize time-continuous automatic sport news reporting.","PeriodicalId":135595,"journal":{"name":"2022 International Conference on Culture-Oriented Science and Technology (CoST)","volume":"73 2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133564425","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-01DOI: 10.1109/cost57098.2022.00078
Nuo Pang, Yan Liu, Yujiang Ji
Nowadays, with the rapid development of science and technology, in the face of the increasingly fast pace of life, the frequent replacement of technologically innovative products, and the intensification of the aging of the domestic population, people’s living habits are constantly changing. People need to spend a lot of learning costs to form new habits and adapt to new technological products. This is tantamount to aggravating the learning pressure of people’s lives and compressing the time that is not sufficient. This paper firstly analyzes and sorts out the core ideas and design ideas of the concept of “unconscious design”. We then analyze the design products of this design concept, investigate and refer to, extract the core ideas and concepts of “unconscious design” applied to product modeling. Finally, we combine the research data, extract modeling elements in the process of product design, and combine the “unconscious design concept”. Data analysis on materials, colors, and human-computer interaction provides real and effective design cases to the research of the subject.
{"title":"Investigation on the application of unconscious design in product modeling","authors":"Nuo Pang, Yan Liu, Yujiang Ji","doi":"10.1109/cost57098.2022.00078","DOIUrl":"https://doi.org/10.1109/cost57098.2022.00078","url":null,"abstract":"Nowadays, with the rapid development of science and technology, in the face of the increasingly fast pace of life, the frequent replacement of technologically innovative products, and the intensification of the aging of the domestic population, people’s living habits are constantly changing. People need to spend a lot of learning costs to form new habits and adapt to new technological products. This is tantamount to aggravating the learning pressure of people’s lives and compressing the time that is not sufficient. This paper firstly analyzes and sorts out the core ideas and design ideas of the concept of “unconscious design”. We then analyze the design products of this design concept, investigate and refer to, extract the core ideas and concepts of “unconscious design” applied to product modeling. Finally, we combine the research data, extract modeling elements in the process of product design, and combine the “unconscious design concept”. Data analysis on materials, colors, and human-computer interaction provides real and effective design cases to the research of the subject.","PeriodicalId":135595,"journal":{"name":"2022 International Conference on Culture-Oriented Science and Technology (CoST)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131093633","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-01DOI: 10.1109/CoST57098.2022.00093
Lili Zhu, Tao Pang
This study aims to propose that a digital platform should be built for the protection of intangible cultural heritage (ICH). The Beijing section of the Great Wall Cultural Belt is used as the study area. It introduces how to establish a resource database and how to build a digital platform for the protection, inheritance, development and promotion of the relevant ICH. It will also develop and design a variety of application platforms for digital terminals, forming an application platform for terminals with regional cultural characteristics, such as information enquiry, intelligent navigation for tourism, appreciation of masterpieces of arts and crafts, and learning and interaction of ICH, so as to realise the ultimate wish of experiencing ICH anytime and anywhere. It also provides interactive designs based on platforms such as human-computer interaction and artificial intelligence to increase the public’s enthusiasm for participation and expand the role and significance of intangible cultural heritage protection, heritage, propaganda and promotion.
{"title":"Research on Digital Platform Technology of Intangible Cultural Heritage in Beijing Section of Great Wall Cultural Belt","authors":"Lili Zhu, Tao Pang","doi":"10.1109/CoST57098.2022.00093","DOIUrl":"https://doi.org/10.1109/CoST57098.2022.00093","url":null,"abstract":"This study aims to propose that a digital platform should be built for the protection of intangible cultural heritage (ICH). The Beijing section of the Great Wall Cultural Belt is used as the study area. It introduces how to establish a resource database and how to build a digital platform for the protection, inheritance, development and promotion of the relevant ICH. It will also develop and design a variety of application platforms for digital terminals, forming an application platform for terminals with regional cultural characteristics, such as information enquiry, intelligent navigation for tourism, appreciation of masterpieces of arts and crafts, and learning and interaction of ICH, so as to realise the ultimate wish of experiencing ICH anytime and anywhere. It also provides interactive designs based on platforms such as human-computer interaction and artificial intelligence to increase the public’s enthusiasm for participation and expand the role and significance of intangible cultural heritage protection, heritage, propaganda and promotion.","PeriodicalId":135595,"journal":{"name":"2022 International Conference on Culture-Oriented Science and Technology (CoST)","volume":"78 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120983029","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-01DOI: 10.1109/cost57098.2022.00077
Wei Chow, M. Shieh, Fan Wu, Han Gao
The study designed the image questionnaires based on nine representative samples and 20 adjective images of Tujia Brocade. Also, the morphological matrix was achieved through the focus group. The morphological matrix includes five features and corresponding feature types of Tujia Brocade. Then, the image of each feature type was calculated based on the multivariate analysis of variance (MANOVA). The image system of Tujia Brocade feature types was constructed. The analysis results showed that five features of Tujia Brocade could have significant influence on the different vocabulary images. This study investigated the relationship between the vocabulary images and the feature types of Tujia Brocade. The study can provide references for designers and relevant manufacturers to understand consumers’ psychology and help Tujia Brocade to update and renovate.
本研究选取9个代表性样本和20个土家织锦的形容词形象设计形象问卷。通过焦点小组获得形态学矩阵。形态矩阵包括土家锦的五种特征和相应的特征类型。然后，基于多变量方差分析(multivariate analysis of variance, MANOVA)对各特征类型的图像进行计算。构建了土家织锦特征类型的图像系统。分析结果表明，土家织锦的五个特征对不同的词汇形象有显著的影响。本研究探讨了土家织锦的词汇意象与特征类型之间的关系。本研究可为设计师及相关厂商了解消费者心理提供参考，帮助土家织锦进行更新改造。
{"title":"A research on the image system of “Tujia Brocade” feature types","authors":"Wei Chow, M. Shieh, Fan Wu, Han Gao","doi":"10.1109/cost57098.2022.00077","DOIUrl":"https://doi.org/10.1109/cost57098.2022.00077","url":null,"abstract":"The study designed the image questionnaires based on nine representative samples and 20 adjective images of Tujia Brocade. Also, the morphological matrix was achieved through the focus group. The morphological matrix includes five features and corresponding feature types of Tujia Brocade. Then, the image of each feature type was calculated based on the multivariate analysis of variance (MANOVA). The image system of Tujia Brocade feature types was constructed. The analysis results showed that five features of Tujia Brocade could have significant influence on the different vocabulary images. This study investigated the relationship between the vocabulary images and the feature types of Tujia Brocade. The study can provide references for designers and relevant manufacturers to understand consumers’ psychology and help Tujia Brocade to update and renovate.","PeriodicalId":135595,"journal":{"name":"2022 International Conference on Culture-Oriented Science and Technology (CoST)","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128717640","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-01DOI: 10.1109/CoST57098.2022.00081
Yi Yu, Yige Gao, Bo Song
The upgrading of media technology brings changes in media forms, and Computer-Mediated Communication (CMC) is constructing the cultural landscape of the new era. Network live stream, which is mainly characterized by technology empowerment, multiple platforms, two-way communication and synchronous interaction, satisfies the sensory experience, interactive experience and emotional experience of the audience. And it further blurs the boundary between virtual and reality. Based on CMC vision, this paper, with the reference of media context theory, aims to explore the psychology and behavior of participants in live-streaming under the state of human-computer interactions. The inner connection among the characteristics of immersive small-screen live-streaming, audience psychology, and behavior would be analyzed. This study attempts to explain the new cultural landscape constructed by online immersive live stream and experience.
{"title":"A Study on the Presentation of Immersive Mobile Live Stream Cultural Landscape from the Perspective of Computer-Mediated Communication","authors":"Yi Yu, Yige Gao, Bo Song","doi":"10.1109/CoST57098.2022.00081","DOIUrl":"https://doi.org/10.1109/CoST57098.2022.00081","url":null,"abstract":"The upgrading of media technology brings changes in media forms, and Computer-Mediated Communication (CMC) is constructing the cultural landscape of the new era. Network live stream, which is mainly characterized by technology empowerment, multiple platforms, two-way communication and synchronous interaction, satisfies the sensory experience, interactive experience and emotional experience of the audience. And it further blurs the boundary between virtual and reality. Based on CMC vision, this paper, with the reference of media context theory, aims to explore the psychology and behavior of participants in live-streaming under the state of human-computer interactions. The inner connection among the characteristics of immersive small-screen live-streaming, audience psychology, and behavior would be analyzed. This study attempts to explain the new cultural landscape constructed by online immersive live stream and experience.","PeriodicalId":135595,"journal":{"name":"2022 International Conference on Culture-Oriented Science and Technology (CoST)","volume":"91 5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126122356","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Digital audio watermarking algorithm is a technology that embeds the watermark signal with special significance into the original audio signal through some method, and then extracts the watermark signal when necessary. It has important applications in the fields of copyright protection, content authentication, secure communication and bill anti-counterfeiting. With the popularity of various multimedia software and audio editing software, it becomes more and more convenient to modify audio. Common audio editing operations such as editing, shifting, tone sandhi and splicing are widely used, but most of the existing digital audio watermarking algorithms can not resist shifting operations. Therefore, this paper proposes an audio watermarking algorithm for audio variable speed attack. The algorithm is based on WSOLA principle. In the process of audio speed change, the spread spectrum watermark is embedded into the non overlapping region.
{"title":"Audio Watermarking Algorithm Based on WSOLA Resisting Speed Change","authors":"Xin Wang, Ying Huang, Shiqiang Wu, Hu Guan, Baoning Niu, Shuwu Zhang","doi":"10.1109/CoST57098.2022.00059","DOIUrl":"https://doi.org/10.1109/CoST57098.2022.00059","url":null,"abstract":"Digital audio watermarking algorithm is a technology that embeds the watermark signal with special significance into the original audio signal through some method, and then extracts the watermark signal when necessary. It has important applications in the fields of copyright protection, content authentication, secure communication and bill anti-counterfeiting. With the popularity of various multimedia software and audio editing software, it becomes more and more convenient to modify audio. Common audio editing operations such as editing, shifting, tone sandhi and splicing are widely used, but most of the existing digital audio watermarking algorithms can not resist shifting operations. Therefore, this paper proposes an audio watermarking algorithm for audio variable speed attack. The algorithm is based on WSOLA principle. In the process of audio speed change, the spread spectrum watermark is embedded into the non overlapping region.","PeriodicalId":135595,"journal":{"name":"2022 International Conference on Culture-Oriented Science and Technology (CoST)","volume":"48 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121675998","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-01DOI: 10.1109/cost57098.2022.00038
Yilan Zhou, Yue Zhang, Shanzhen Lan, Shaobin Li
The glass-free three dimensional (3D) large screen has provided an ultra-sensory immersive visual experience, and broken the aesthetic fatigue of people to print advertising. But it’s also limited by the geographical location and high cost of installation, accordingly, the virtual large curved screens on multiple buildings with interaction effect based on mixed reality (MR) was designed and implemented in this paper. Firstly, the deep learning building detection and building surface recognition algorithm were adopted to recognize the building surfaces, and the recognized surfaces were further filtered and integrated, which then completed the positioning of multiple buildings virtual large screens based on the integrated multiple plane groups. Based on the simultaneous rendering of multiple virtual curved screens, several interaction modes were also designed between the large screen spaces, thus furtherly enhancing the user experience and application potential of the virtual curved screens. The overall program built modularly based on 3D engine achieves better recognition, rendering and interaction effects in mobile, which also provides new ideas for new forms of MR applications in urban environments.
{"title":"Implementation of the Interaction Effect Among Virtual Large Curved Screens on Multiple Buildings Based on Mixed Reality","authors":"Yilan Zhou, Yue Zhang, Shanzhen Lan, Shaobin Li","doi":"10.1109/cost57098.2022.00038","DOIUrl":"https://doi.org/10.1109/cost57098.2022.00038","url":null,"abstract":"The glass-free three dimensional (3D) large screen has provided an ultra-sensory immersive visual experience, and broken the aesthetic fatigue of people to print advertising. But it’s also limited by the geographical location and high cost of installation, accordingly, the virtual large curved screens on multiple buildings with interaction effect based on mixed reality (MR) was designed and implemented in this paper. Firstly, the deep learning building detection and building surface recognition algorithm were adopted to recognize the building surfaces, and the recognized surfaces were further filtered and integrated, which then completed the positioning of multiple buildings virtual large screens based on the integrated multiple plane groups. Based on the simultaneous rendering of multiple virtual curved screens, several interaction modes were also designed between the large screen spaces, thus furtherly enhancing the user experience and application potential of the virtual curved screens. The overall program built modularly based on 3D engine achieves better recognition, rendering and interaction effects in mobile, which also provides new ideas for new forms of MR applications in urban environments.","PeriodicalId":135595,"journal":{"name":"2022 International Conference on Culture-Oriented Science and Technology (CoST)","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128205468","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-08-01DOI: 10.1109/CoST57098.2022.00015
Hongyang Zhou, Yi Ma, Yan Ma, Xiaobin Zhu
Most existing image super-resolution (SR) methods has achieved superior performance. However, the contrastive learning, which commonly be used in high-level tasks, has not been fully exploited in existing deep learning based image SR methods. In this paper, we propose an image super-resolution network based feedback mechanism to learn abstract representations to explore high information in the representation space. Specifically, we first use the hidden states and constraints in RNN to achieve feedback network. Then, a contrastive learning is used to conduct representation learning by pulling the final SR image to the high resolution image and push the final image to intermediate images. In addition, we introduce a split based feedback block (SPFB) to reduce the redundancy of models for inference acceleration, where the tolerate features with similar patterns but require less computation. Extensive experimental results demonstrate the superiority of the proposed method in comparison with the state-of-the-art methods.
{"title":"Split-based Feedback Network for Image Super-Resolution","authors":"Hongyang Zhou, Yi Ma, Yan Ma, Xiaobin Zhu","doi":"10.1109/CoST57098.2022.00015","DOIUrl":"https://doi.org/10.1109/CoST57098.2022.00015","url":null,"abstract":"Most existing image super-resolution (SR) methods has achieved superior performance. However, the contrastive learning, which commonly be used in high-level tasks, has not been fully exploited in existing deep learning based image SR methods. In this paper, we propose an image super-resolution network based feedback mechanism to learn abstract representations to explore high information in the representation space. Specifically, we first use the hidden states and constraints in RNN to achieve feedback network. Then, a contrastive learning is used to conduct representation learning by pulling the final SR image to the high resolution image and push the final image to intermediate images. In addition, we introduce a split based feedback block (SPFB) to reduce the redundancy of models for inference acceleration, where the tolerate features with similar patterns but require less computation. Extensive experimental results demonstrate the superiority of the proposed method in comparison with the state-of-the-art methods.","PeriodicalId":135595,"journal":{"name":"2022 International Conference on Culture-Oriented Science and Technology (CoST)","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127639105","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：