Sheng Jin
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Sheng Jin has a total of 9 co-authors. Here are the top co-authors by number of shared publications:
- Huaizhong Lin
College of Computer Science and Technology, Zhejiang University, Hangzhou 310027, China
1 shared publication - Lei Zhao
College of Computer Science and Technology, Zhejiang University, Hangzhou 310027, China
1 shared publication - Lixia Chen
School of Humanities and Communication, Zhejiang Gongshang University, Jiaogong Road, Hangzhou 310018, China
1 shared publication - Min Fan
Communication University of China, Beijing, China
1 shared publication - Tianyi Chu
College of Computer Science and Technology, Zhejiang University, No. 38 Zhejiang Road, Hangzhou 310027, China
1 shared publication - Wei Xing
College of Computer Science and Technology, Zhejiang University, Hangzhou, China
1 shared publication
Latest Publications
- Open Access
Designing an Accountable Generative AI Co-Creation System for Museum Visitors: A Field Study at the Lanting Calligraphy Museum
Heritage museums increasingly seek to support visitors as active meaning-makers during the visit and to extend that experience beyond the museum through personally meaningful takeaways. Generative AI (GenAI) may support this goal by helping visitors transform museum source material into personalized artifacts, but open-ended generation may weaken heritage linkage and obscure curatorial responsibility. This paper proposes two paired design principles for heritage GenAI: bounded creativity, which defines what visitors may change, and curatorial accountability, which defines how cultural limits are set, explained, and revised. We implemented these principles at the Lanting Calligraphy Museum in Shaoxing, China, through a visit-linked co-creation system. Visitors collected source images during the museum tour and later created a personalized 3D-printed takeaway through either a rule-based editor with predefined templates and options or a GenAI-mediated dialogue interface. A between-subjects field study with adult visitors (N = 60; 20 per condition) compared conventional consumption (a standard tour and souvenir), rule-based co-creation, and GenAI-mediated co-creation. Both co-creation pathways improved cultural engagement, perceived connection to heritage sources, personal meaning and self-expression, and psychological ownership compared with conventional consumption. The GenAI-mediated pathway produced further gains in cultural engagement, personal meaning, self-expression, and psychological ownership. Still, it did not reliably increase perceived connection to heritage sources beyond that of the rule-based pathway. These findings suggest that the GenAI pathway added value mainly on personal authenticity and psychological ownership, by helping visitors articulate and revise intentions within accountable cultural boundaries; its incremental effect on heritage-linked authenticity remains undetermined at this sample size.
Open AccessDesigning an Accountable Generative AI Co-Creation System for Museum Visitors: A Field Study at the Lanting Calligraphy Museum
Heritage museums increasingly seek to support visitors as active meaning-makers during the visit and to extend that experience beyond the museum through personally meaningful takeaways. Generative AI (GenAI) may support this goal by helping visitors transform museum source material into personalized artifacts, but open-ended generation may weaken heritage linkage and obscure curatorial responsibility. This paper proposes two paired design principles for heritage GenAI: bounded creativity, which defines what visitors may change, and curatorial accountability, which defines how cultural limits are set, explained, and revised. We implemented these principles at the Lanting Calligraphy Museum in Shaoxing, China, through a visit-linked co-creation system. Visitors collected source images during the museum tour and later created a personalized 3D-printed takeaway through either a rule-based editor with predefined templates and options or a GenAI-mediated dialogue interface. A between-subjects field study with adult visitors (N = 60; 20 per condition) compared conventional consumption (a standard tour and souvenir), rule-based co-creation, and GenAI-mediated co-creation. Both co-creation pathways improved cultural engagement, perceived connection to heritage sources, personal meaning and self-expression, and psychological ownership compared with conventional consumption. The GenAI-mediated pathway produced further gains in cultural engagement, personal meaning, self-expression, and psychological ownership. Still, it did not reliably increase perceived connection to heritage sources beyond that of the rule-based pathway. These findings suggest that the GenAI pathway added value mainly on personal authenticity and psychological ownership, by helping visitors articulate and revise intentions within accountable cultural boundaries; its incremental effect on heritage-linked authenticity remains undetermined at this sample size.
- Open Access
- 19 March 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (6) , 2969
CSCGAN: Cross-Space Contrastive Learning for Blind Image Inpainting
Existing general image inpainting works require the user to customize a mask to indicate the region to be inpainted. However, the mask is often hard to calibrate accurately in real-world applications, e.g., graffiti removal. Blind image inpainting aims to automatically restore the degraded image into the visually reasonable one without a priori mask to indicate the area to be repaired. So far, most proposed blind inpainting methods convert the task into general inpainting by predicting the mask before inpainting. However, these methods are highly dependent on mask prediction results, which may produce inferior inpainting results if the prediction is inaccurate. To address this issue, we propose a two-stage blind inpainting framework with two novel designs: (1) cross-space contrastive learning, to remove the noise in the degraded images and realize the automatic inpainting in the latent space by reducing the distance of the degraded images and the corresponding complete images in the latent space; and (2) mask-aware adversarial training, to minimize the mutual information between the inpainted feature and the noise. Extensive experiments prove that our blind inpainting framework performs better on multiple datasets than the state-of-the-art methods.
- 19 March 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (6) , 2969
Open AccessCSCGAN: Cross-Space Contrastive Learning for Blind Image Inpainting
Existing general image inpainting works require the user to customize a mask to indicate the region to be inpainted. However, the mask is often hard to calibrate accurately in real-world applications, e.g., graffiti removal. Blind image inpainting aims to automatically restore the degraded image into the visually reasonable one without a priori mask to indicate the area to be repaired. So far, most proposed blind inpainting methods convert the task into general inpainting by predicting the mask before inpainting. However, these methods are highly dependent on mask prediction results, which may produce inferior inpainting results if the prediction is inaccurate. To address this issue, we propose a two-stage blind inpainting framework with two novel designs: (1) cross-space contrastive learning, to remove the noise in the degraded images and realize the automatic inpainting in the latent space by reducing the distance of the degraded images and the corresponding complete images in the latent space; and (2) mask-aware adversarial training, to minimize the mutual information between the inpainted feature and the noise. Extensive experiments prove that our blind inpainting framework performs better on multiple datasets than the state-of-the-art methods.