🧐 About Me
Hi there! I am a master student in Computer Science at the Fudan University, under the supervision of Prof. Yu-Gang Jiang and Prof. Zhineng Chen. Before that, I completed my Bachelor’s degree in Wuhan University of Science and Technology in July 2021.
Research Interests: I primarily focus on the cross-modal representation and alignment of visual-language models in text recognition 📚. Specifically, my research involves the integration of semantics and visual cues in scene text, modeling image features and character positions in text correction, and the fusion of language expertise in continuous multilingual texts.
My personal tech blog has surpassed 210,000 views! If you’re interested in it, please click here 📃.
🔥 News
- 2023.10: 🥳 I have received the graduate national scholarship with a top-ranked (1/380) overall performance!
- 2023.08: 🎉 One papers is accepted by IJCV 2023 (first author)!
- 2023.07: 🎉 One papers is accepted by ICCV 2023 (first author)!
- 2023.05: 🎉 One papers is accepted by IJCAI 2023 (first author)!
- 2022.05: One papers is accepted by IJCAI 2022!
- 2020.12: 🥳 I have received the undergraduate national scholarship with a top-ranked (1/236) overall performance!
- 2019.08: ⚽ My team won 2nd place in Simurosot Large Size (1st person) at 24th FIRA RoboWorld Cup!
- 2019.08: ⚽ My team won 3rd place in Simurosot Middle Size (3rd person) at 24th FIRA RoboWorld Cup!
📝 Publications
CDistNet: Perceiving Multi-Domain Character Distance for Robust Text Recognition
Tianlun Zheng, Zhineng Chen*, Shancheng Fang, Hongtao Xie, Yu-Gang Jiang.
International Journal of Computer Vision (IJCV 2023)
Code
PDF
- The lack of alignment between visual and semantic aspects in attention can lead to attention drift. Positional attention lacks learnable positional supervision. CDistNet incorporates additional learnable character position branches to separately query semantic and visual cues, thereby enhancing character position alignment and the integration of visual-semantic cues.
MRN: a Multiplexed Routing Network for Multilingual Incremental Text Recognition
Tianlun Zheng, Zhineng Chen*, BingChen Huang, Wei Zhang, Yu-Gang Jiang.
International Conference on Computer Vision (ICCV 2023)
Code
PDF
- This is the first work that combines incremental learning with scene text. Unlike other incremental learning scenarios, scene text exhibits unique characteristics, leading to the rehearsal-imbalance issue. To address this, we propose the Multiplexed Routing Network, which reduces reliance on the rehearsal set. Compared to existing state-of-the-art methods, our approach achieves accuracy improvements ranging from 10.3% to 27.4%.
TPS++: Attention-Enhanced Thin-Plate Spline for Scene Text Recognition
Tianlun Zheng, Zhineng Chen*, Jinfeng Bai, Hongtao Xie, Yu-Gang Jiang.
International Joint Conferences on Artificial Intelligence (IJCAI 2023)
Code
PDF
- TPS++ is the first work to introduce the attention mechanism into text rectification. Traditional text rectification methods lack modifications to the rectification transformation formula, which can lead to unnatural deformations and character out-of-bounds issues. TPS++ introduce an attention mechanism to provide greater flexibility in rectification and shift the rectification process from the image level to the feature level.
- IJCV 2023 CDistNet: Perceiving Multi-Domain Character Distance for Robust Text Recognition, Tianlun Zheng, Zhineng Chen, Shancheng Fang, Hongtao Xie, Yu-Gang Jiang.
- ICCV 2023 MRN: a Multiplexed Routing Network for Multilingual Incremental Text Recognition, Tianlun Zheng, Zhineng Chen, BingChen Huang, Wei Zhang, Yu-Gang Jiang.
- IJCAI 2023 TPS++: Attention-Enhanced Thin-Plate Spline for Scene Text Recognition, Tianlun Zheng, Zhineng Chen, Jinfeng Bai, Hongtao Xie, Yu-Gang Jiang.
- IJCAI 2022 Oral SVTR: Scene Text Recognition with a Single Visual Model, Yongkun Du, Caiyan Jia, Zhineng Chen, Xiaoting Yin, Tianlun Zheng, Chenxia Li, Yuning Du, Yu-Gang Jiang.
- ICIEA 2020 A Novel Method based on Character Segmentation for Slant Chinese Screen-render Text Detection and Recognition, Tianlun Zheng, Xiaofeng Wang, Xin Yuan, Shiqin Wang.
💾 Invention Patent
- “A Method and Apparatus for Image Text Correction Based on Attention-Enhanced Thin Plate Spline Transformation,” Application Number 202310536598.5
- “A Method for Scene Text Recognition Based on Character Distance Perception,” Application Number 202210689812.6
- “A Badminton Motion Analysis System and Method,” Application Number 201910981141.9
- “Machine Vision-Based Planar Four-Bar Linkage Mechanism Pose Detection System and Detection Method,” Granted
🌍 Competitions
- 🤖 Robotics
- 2nd place in Simurosot Large Size at 24th FIRA RoboWorld Cup, Changwon, Korea, August 2019.
- 3rd place in Simurosot Middle Size at 24th FIRA RoboWorld Cup, Changwon, Korea, August 2019.
- First Prize in the Robot Racing Obstacle Run Category at the 20th China Robot and Artificial Intelligence Contest, October 2018.
- Second Prize in the Robot Racing Sprint Category at the 20th China Robot and Artificial Intelligence Contest, October 2018.
- First Prize in the 5v5 Category at the 21st China Robot and Artificial Intelligence Contest, October 2019.
- Third Prize in the 11v11 Category at the 21st China Robot and Artificial Intelligence Contest, October 2019.
- Second Prize in the Human Identification Category at the China Service Robot Competition, May 2018.
- First Prize in the Second “Longteng Cup” National College Student Creative and Innovation Contest, November 2019.
- 🎨 Computer Vision
- 5th (5 / 1364) place in the Lightweight Text Recognition Academic Competition organized by the Chinese Society of Graphics and Image.
The program was promoted by Baidu Official Account Link, August 2021. - 4th place in the Greater Bay Area (Huangpu) International Algorithm Instance Competition - Street View Image Shop Sign Text Recognition Competition, November 2022.
- 5th (5 / 1364) place in the Lightweight Text Recognition Academic Competition organized by the Chinese Society of Graphics and Image.
🎖 Honors and Student Services
- 🏅Honors
-
2017-2018: School Outstanding Scholarship (1%), Xu Jiayin Scholarship (0.2%), Excellent Student Leader.
-
2018-2019: School Outstanding Scholarship (1%), Xu Jiayin Scholarship (0.2%), Exemplary Student Model.
-
2019-2021: School Outstanding Scholarship (1%), National Scholarship (0.2%), Exemplary Student Model.
-
2021-2022: Outstanding Graduate, Excellent Recommended Student Scholarship (5%).
-
2022-2023: Outstanding Communist Youth League Member.
-
- 👨🎓 Student Services
- 2017-2018: Class Monitor
- 2018-2019: Class Academic Representative, Vice President of School Guitar Association.
- 2019-2020: Class Academic Representative, Vice President of the School’s Simulated Robot Soccer Association.
📖 Educations
- 🎓 2021.09 - 2024.02, Master, Fudan University, Shanghai.
- 🎓 2017.09 - 2021.06, Undergraduate, Wuhan University of Science and Technology, Wuhan.
💬 Academic Services
- Conference Reviewer:
- ACM International Conference on Multimedia (ACM MM 2021-2023)
- AAAI Conference on Artificial Intelligence (AAAI 2022)
- The IEEE International Conference on Multimedia & Expo (ICME 2022-2023)
- Journal Reviewer:
- IEEE Transactions on Circuits and Systems for Video Technology
💻 Internships
- 2020.10 - 2021.02, ByteDance, Recommendation Algorithms Research Intern, Beijing.
- 2021.10 - 2021.12, Baidu, Computer Vision Research Intern, Beijing.
- 2023.05 - 2023.09, Alibaba Eleme, Recommendation Algorithms Research Intern, Shanghai.