Publications & Projects

Some selected publications & personal projects.

Image Re-ranking with Long-Context Sequence Modeling
Zilin Xiao, Ayush Sachdeva, Hao-Jen Wang and Vicente Ordonez
In the Instance-Level Recognition Workshop at ECCV'24 (Short Paper)
[PDF]

Will the Real Linda Please Stand up...to Large Language Models? Examining the Representativeness Heuristic in LLMs
Pengda Wang1, Zilin Xiao1, Hanjie Chen and Frederick L. Oswald.
In the 2024 Conference on Language Modeling (COLM'24) as Oral Spotlight
[arxiv] [Code]

Grounding Language Models for Visual Entity Recognition
Zilin Xiao, Ming Gong, Paola Cascante-Bonilla, Xingyao Zhang, Jie Wu and Vicente Ordonez.
In the 2024 European Conference on Computer Vision (ECCV'24)
[Code] [arxiv]

Instructed Language Models with Retriever Are Powerful Entity Linkers
Zilin Xiao, Ming Gong, Jie Wu, Xingyao Zhang, Linjun Shou and Daxin Jiang.
In the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP'23)
[Code] [PDF] [aclanthology] [Poster]

Coherent Entity Disambiguation via Modeling Topic and Categorical Dependency
Zilin Xiao, Linjun Shou, Xingyao Zhang, Jie Wu, Ming Gong and Daxin Jiang.
In the 2023 Findings of Empirical Methods in Natural Language Processing (Findings of EMNLP'23)
[PDF] [aclanthology]

Exploring the Effectiveness of Appearance Descriptor in DeepSORT
Zilin Xiao and Yanan Sun.
In the 2022 International Joint Conference on Neural Networks (IJCNN'22)
[Code] [PDF]


openai-managerGitHub stars PyPI status
Speed up your OpenAI requests by balancing prompts to multiple API keys.
[Blog Post (in Chinese)] [Code] [pypi.org]

Hyper-Table-OCRGitHub stars
A carefully-designed OCR pipeline for universal boarded table recognition and reconstruction.
[Blog Post (in Chinese)] [Code] [Video] [Video (in Mandarin)]

Auto-Make-MoneyGitHub stars
An easy tool to help investors from Shanghai/Shenzhen Stock Market to apply for new-listed convertible bonds automatically.
[Blog Post (in Chinese)] [Code]

DroneObjectDetectionGitHub stars
A UAV object detection system based on YOLOv3, providing an interactive interface and `VisDrone2018` training scheme.
[Code] [Dataset (Passcode: kdfv)]

1. Equal Contribution.