Junyu Chen portrait

个人主页 Personal Site

Junyu Chen

人工智能专业 · 广州航海学院 / 广州交通大学（筹） Artificial Intelligence Major · Guangzhou Maritime University / Guangzhou Jiaotong University (Proposed)

我目前就读于广州航海学院 / 广州交通大学（筹）人工智能专业，研究兴趣主要集中在人工智能、 工业异常检测与 AI 视觉方向。在研究与学习中，我重视严谨的实验过程、可靠的结果验证， 也始终把科研精神与学术道德放在非常重要的位置。

I am currently studying Artificial Intelligence at Guangzhou Maritime University / Guangzhou Jiaotong University (Proposed). My interests focus on artificial intelligence, industrial anomaly detection, and AI vision. In both research and engineering practice, I value rigorous experimentation, reliable validation, and a serious commitment to academic integrity.

研究兴趣：人工智能 · 多模态 · 工业异常检测 · AI视觉 Research Interests: AI · Multimodal Learning · Industrial Anomaly Detection · AI Vision
查看项目 View Projects 联系我 Contact Me

关于我 About Me

我目前就读于广州航海学院 / 广州交通大学（筹）人工智能专业，研究兴趣主要集中在人工智能、 工业异常检测与 AI 视觉方向。在深圳大学博士后兼硕士研究生导师曾德宇老师的指导下， 我持续关注如何将算法研究与实际工业场景结合起来，并努力在问题定义、实验设计、结果分析与复现规范上保持严谨。 我希望通过长期积累，形成兼具研究深度与工程落地能力的技术路径，同时始终坚持精益求精的科研精神与严肃认真的学术道德。

I am currently studying Artificial Intelligence at Guangzhou Maritime University / Guangzhou Jiaotong University (Proposed), with research interests in artificial intelligence, multimodal learning, industrial anomaly detection, and AI vision. Under the guidance of Dr. Deyu Zeng, a postdoctoral researcher at Shenzhen University and a master's supervisor, I keep exploring how algorithmic research can connect with real industrial scenarios while maintaining rigor in problem formulation, experimental design, result analysis, and reproducibility. I hope to develop both solid research depth and practical engineering ability, while always treating academic integrity and a meticulous research attitude as core personal principles.

技术关键词 Keywords

Artificial Intelligence Multimodal Learning Industrial Anomaly Detection AI Vision PyTorch OpenCV Model Deployment

目标 Goal

  • 持续保持精益求精的科研精神与严谨端正的学术道德。
  • 在人工智能、多模态、工业异常检测与 AI 视觉方向继续积累研究与工程能力。
  • 未来三年后报考深圳大学机电学院硕士研究生。
  • To maintain a meticulous research attitude and strong academic integrity in all future work.
  • To continue building research and engineering ability in AI, multimodal learning, industrial anomaly detection, and AI vision.
  • To apply for a master's program at the College of Mechatronics and Control Engineering, Shenzhen University, in about three years.

动态 News

这里记录近期的重要更新、论文录用、项目进展与阶段性成果。

This section highlights recent updates, paper acceptances, project milestones, and notable progress.

论文录用 Paper Accepted 2026.06

一篇论文被欧洲计算机视觉会议（ECCV 2026）录用！

One paper was accepted by the European Conference on Computer Vision (ECCV 2026).

这是当前主页上的首条正式动态，后续可以继续在这里追加论文、竞赛、项目里程碑和其他重要更新。

This is the first formal news item on the homepage, and more updates such as papers, competitions, project milestones, or other key achievements can be added here later.

论文 Selected Publications

这里用于展示已发表或已录用的代表性论文。后续补充论文题目、作者信息、摘要与链接后，这一部分会更完整。

This section presents selected accepted or published papers. It can be expanded with titles, author lists, abstracts, and paper links as more details become available.

2026
ECCV 2026 Accepted · 2026.06

CMDS-AD: Cross-Modal Dual-Stream Decoupling for Few-Shot Anomaly Detection

CMDS-AD: Cross-Modal Dual-Stream Decoupling for Few-Shot Anomaly Detection

Junhao Cai, Junyu Chen, Deyu Zeng, Junhao Pang, Qiwei Liang, Xiaopin Zhong, Zongze Wu

Junhao Cai, Junyu Chen, Deyu Zeng, Junhao Pang, Qiwei Liang, Xiaopin Zhong, Zongze Wu

European Conference on Computer Vision (ECCV), 2026

该工作面向小样本异常检测场景，提出跨模态双流解耦框架，结合 RGB 与 3D 几何信息缓解训练样本稀缺带来的性能瓶颈， 并通过更精细的跨模态对齐与噪声抑制提升异常检测效果。

This work targets few-shot anomaly detection by proposing a cross-modal dual-stream decoupling framework that leverages both RGB cues and 3D geometric information to overcome severe data scarcity while improving cross-modal alignment and anomaly localization.

Code Paper: Coming Soon Paper: Coming Soon

项目展示 Selected Projects

这里展示我目前最想重点呈现的研究项目、方向性工作与代表性成果。

This section highlights the research project, direction-specific work, and representative achievement I want to showcase most.

Featured Project ECCV 2026
CMDS-AD: Cross-Modal Dual-Stream Decoupling for Few-Shot Anomaly Detection CMDS-AD: Cross-Modal Dual-Stream Decoupling for Few-Shot Anomaly Detection

这是一个面向小样本异常检测的跨模态工作，结合 RGB 与 3D 几何信息，通过双流解耦设计、 跨模态语义对齐与噪声抑制机制，提升极少样本条件下的异常检测性能。

This project focuses on few-shot anomaly detection by leveraging both RGB and 3D geometric cues. It introduces a dual-stream decoupling design, cross-modal semantic alignment, and noise suppression to improve anomaly detection under extremely limited data settings.

作者：Junhao Cai, Junyu Chen, Deyu Zeng, Junhao Pang, Qiwei Liang, Xiaopin Zhong, Zongze Wu

Authors: Junhao Cai, Junyu Chen, Deyu Zeng, Junhao Pang, Qiwei Liang, Xiaopin Zhong, Zongze Wu

类型 Type 研究项目 Research Project
方向 Focus 多模态异常检测 Multimodal Anomaly Detection
状态 Status ECCV 2026 录用 Accepted by ECCV 2026
Code 查看论文 View Publication
Research Direction
工业异常检测相关项目 Industrial Anomaly Detection Work

这里将继续整理我在工业异常检测、多模态感知、AI 视觉与模型落地相关方向上的后续项目和阶段性工作。

This card will continue to gather my follow-up work in industrial anomaly detection, multimodal perception, AI vision, and practical model deployment.

后续更新 Update Later
Honors & Awards
荣誉奖励 Honors & Awards

2026 年广东省大学生计算机设计大赛工业互联网技术应用赛（本科组）省级决赛三等奖。

Third Prize in the Provincial Final of the 2026 Guangdong College Student Computer Design Competition, Industrial Internet Technology Application Track (Undergraduate Division).

2026 · 省级决赛 2026 · Provincial Final

经历 Experience

当前 Present
学术指导 Academic Guidance 导师：深圳大学博士后兼硕士研究生导师 曾德宇 Mentor: Dr. Deyu Zeng, Shenzhen University postdoctoral researcher and master's supervisor

在曾德宇老师的指导下，我更加重视科研规范、问题意识、实验严谨性与结果可信度， 也在不断训练自己将研究思路转化为可落地的工程实践。

Under Dr. Deyu Zeng's guidance, I place growing emphasis on research discipline, problem awareness, experimental rigor, and trustworthy results, while also learning how to turn research ideas into practical engineering work.

教育 Education
广州航海学院 / 广州交通大学（筹） · 人工智能专业 Guangzhou Maritime University / Guangzhou Jiaotong University (Proposed) · Artificial Intelligence 研究兴趣：人工智能、多模态、工业异常检测、AI视觉 Research interests: AI, multimodal learning, industrial anomaly detection, and AI vision

我希望在现阶段继续夯实人工智能相关基础，积累更扎实的研究与工程训练， 并在未来三年后报考深圳大学机电学院硕士研究生。

At this stage, I hope to keep strengthening my foundations in artificial intelligence, build stronger research and engineering experience, and apply for a master's program at Shenzhen University in about three years.

联系我 Contact

如果你想交流人工智能、多模态、工业异常检测、AI视觉方向，或联系合作与学习机会，可以通过下面的信息找到我。

If you would like to discuss artificial intelligence, multimodal learning, industrial anomaly detection, AI vision, or potential collaboration and study opportunities, feel free to reach out here.

Email GitHub WeChat: CjyCzm1874Cyx

已接入中英文切换、基础信息与头像。后续补充简介和经历后，这个主页就能更完整上线。 The bilingual switcher, photo, and core profile info are in place. Once the summary and experience are added, the homepage will be ready to publish more fully.