About Me
I am a computer vision algorithm engineer in ByteDance.Inc. I’m mainly responsible for multi-modal research, as well as feature optimization and performance improvement of advertising systems.
I received the master degree from Peking University, China under the supervsion of Prof. Yadong Mu. And I’m majored in computer vision and deep learning during that time. My research work mainly involves in Cross modal retrieval, Deep Hashing, Medical AI, Steganography, and Image Captioning.
Before that, I received my bachelor’s degree in computer science from the School of Computer Science in Wuhan University, Wuhan China.
News
One paper acctpted by CVPR 2023.
Learning Instance-Level Representation for Large-Scale Multi-Modal Pretraining in E-commerce
One paper accepted by NeurIPS 2022.
Embracing Consistency: A One-Stage Approach for Spatio-Temporal Video Grounding
One paper accepted by IJCAI 2021.
Research Area
- Computer Vision
- Deep Learning
- Cross-modal Retrieval/Matching
Experiences
Computer Vision Algorithm Engineer in AILab, ByteDance.Inc. (2021.7 - NAN)
- I am working on the application and research of large-scale language models(LLMs) in the commercial field, as well as the construction and evaluation of large-model systems. Which is primarily used for business scenarios such as advertising scripts, live broadcast copywriting, intelligent customer service or product titles generation, and more.
- Previously, I have worked on visual search system in video scenario, the system is widely applied in e-commerce products retrieval as well as advertisements recall and recommendation.
- Responsible for multi-modal research, as well as feature optimization and performance improvement of advertising systems.
Algorithm Engineer Intern in New Retail Intelligent Engine Business Group, Alibaba Group. (2020.6-2020.9)
- Mainly responsible for Object Matting, Advertising Image Generation, Video Understanding.
Winter Camp Intern in Google, China. (2020.1-2020.2)
- Completed an application that can transfer the human in a selfie into animation style and change the background at the same time. It mainly includes three modules: Human Matting, Face style translation and Background neural style transfer. It can synthesized the fine-grained animation human face into the style transferred background image to get a cartoonization photo.
- Completed a program which can achieve real-time character foreground matting and background style transfer in video stream on Intel-i5 CPU.
Research Intern in Medical AI Lab, Tencent. (2018.7-2019.7)
- Responsible for the research and development of the core module in the intelligent medical qualification examination system. The final model can score more than 390 points in the national medical qualification examination, which outperformed more than 70% of the human examinees.
- Researched the automatic generation of medical diagnostic reports for chest X-rays.
- Engeering Intern in Tencent Cloud. (2017.7-2017.9)
- Responsible for the Panshi system decoupling and reconstruction. Separated the system’s web access layer from its data access layer; completed the modification of related data interfaces.
- Developed and maintained the related functional plugins of the system.
- Exchange Student in Hong Kong Baptist University. (2017.5-2017.6)
- Researched on blockchain technology.
Contact
Email: yongzhili at pku.edu.cn
Wechat: lyz327067530
Adress: Someplace, Beijing, China .
Welcome
