Jiahao LI (Jarvis Lee)

Ph.D. Student

City University of Hong Kong (CityUHK)

Email  |  CV (EN)  |  CV (CN)  |  Scholar  |  Github  |  ORCID

About Me

I am currently a (first-year) second-year Ph.D. Student at City University of Hong Kong (CityUHK), supervised by Prof. Jianping Wang. My current research direction is Autonomous Driving, especially 3D Occupancy Prediction, 3D Scene Reconstruction (Depth Estimation & Stereo Matching).

Prior to that, I obtained the Master's Degree of Science in Computer Science from The Chinese University of Hong Kong (CUHK) in 2022, supervised by Prof. Kin Hong Wong and the Bachelor's Degree of Science (Honours) in Computer Science and Technology (First Class) from Beijing Normal - Hong Kong Baptist University (BNBU) and Hong Kong Baptist University (HKBU) in 2021, supervised by Prof. Hui Zhang, Prof. Zhiyuan LI and Prof. Raymond Lee. After graduating from CUHK, I also spent a wonderful time from 2022 to 2024, working at the PICO MR Team of ByteDance Ltd. as a Computer Vision Algorithm Engineer focused on Depth Estimation and 3D Reconstruction, mentored by Xiao Liu.

News

Publications

 Global Regulation and Excitation via Attention Tuning for Stereo Matching
Jiahao LI, Xinhong Chen, Zhengmin JIANG, Qian Zhou, Yung-Hui Li, Jianping Wang
ICCV 2025
Paper | Video | Code | Cite

We propose a universal framework (GREAT) that can be seamlessly integrated into existing iterative stereo matching methods to optimize the ill-posed regions. The GREAT framework employs three well-designed attention modules named Spatial Attention (SA), Matching Attention (MA), and Volume Attention (VA) to extract and excite the global geometric contextual information into the constructed cost volume.

Previous Publications

 Application of YOLOv5 for mask detection on IoT
Cheng Huang, Yishen Liu, Jiahao LI, Hao Tian, Haoyi Chen
Applied and Computational Engineering 2023
Paper | Code | Cite

Improved YOLOv5, a lightweight deep learning model to assist people in mask detection, vehicle counting, and target tracking, which does not take up too many computing resources and requires less storage space and is easier to deploy.

Chaotic Bi-LSTM and Attention HLCO Predictor Based Quantum Price Level Fuzzy Logic Trading System
Jiahao LI, Zihao Huang, Lirong Lin, Yuchen Guo, Raymond Lee
Soft Computing 2023
Paper | Code | Cite

We propose a neural network-guided price predictor using Attention, BiLSTM, Chaotic Neuro-Oscillator, and Quantum Price Level to forecast HLCO prices and a fuzzy logic-based trading strategy is also introduced to overcome fixed order boundaries and limitations of traditional indicators.

Projects

 The 17th F1TENTH Grand Prix (HongKong)
Jinghuai Deng, Bingyuan Huang, Xiaoyun Dong, Hua Hu, Jiahao LI, Jianping Wang
CPS-IoT Week 2024 (Championship)
Project Page

Supervised by Prof. Jianping Wang, our team (FSM Speed) won the championship at the 17th F1TENTH Autonomous Grand Prix during the CPS-IoT Week 2024 in Hong Kong.

Adaptive Error Aware Cost Volume for Stereo Matching
Jiahao LI, Zhengxin Li, Yiping Bao, Guangyuan Zhou, Qiang Rao, Xiao Liu
ECCV 2024 Submission ID 4123
Draft | Code

We propose a dynamic sampling strategy based on an error map to accelerate iteration and a noise-filtering cost volume method for improved disparity prediction accuracy in ill-posed regions.

Image Caption with Full-Transformer
Jiahao LI, Kin Hong Wong
Master Final Thesis in CUHK
Thesis | Code

This is a full attention-based Neural Image Captioning architecture with BERT and Vision-Outlooker.

Complex Networks and the Applications in Deep Learning
Jiahao LI, Xiangying Wei, Zhiyuan LI
Bachelor Final Thesis in BNBU
Thesis | Code

A method to investigate the relationship between topology and performance of the deep learning neural network based on random wired neural network model (RandWiredNN) and a training method (Wired-Weight Training).

Live Long and Prosper: How Herring and Mackerel Affect Scottish Fisheries
Jiahao LI, Yutong Du, Lirong Lin, Jingjin Wu
Mathematical Contest In Modeling 2020 (Meritorious Winner)
Thesis

This thesis investigates the impact of climate change on the migration patterns of herring and mackerel in Scottish fisheries over 50 years. Using linear regression models, the study predicts future sea surface temperatures and identifies optimal habitats for these species. Economic models analyze profit dynamics for small fishing companies, highlighting risks from prolonged transport times and spoilage.

Education

 City University of Hong Kong (CityUHK)
Ph.D. Student in Computer Science      • Sep. 2024 - Present
Supervisor: Prof. Jianping Wang
Research Direction: Autonomous Driving and 3D Scene Reconstruction
The Chinese University of Hong Kong (CUHK)
Master Degree in Computer Science      • Sep. 2021 - Nov. 2022
Supervisor: Prof. Kin Hong Wong
Research Direction: 2D Object Detection and Image Captioning
Hong Kong Baptist University (HKBU)
Bachelor Degree in Computer Science and Technology      • Sep. 2017 - Nov. 2021
Supervisor: Prof. Hui Zhang, Prof. Zhiyuan LI and Prof. Raymond Lee
Research Direction: Deep Learning, Quantum Finance and Graph Neural Network
Beijing Normal - Hong Kong Baptist University (BNBU)
Bachelor Degree in Computer Science and Technology      • Sep. 2017 - Nov. 2021
Supervisor: Prof. Hui Zhang, Prof. Zhiyuan LI and Prof. Raymond Lee
Research Direction: Deep Learning, Quantum Finance and Graph Neural Network

Internship and Full-Time Work

 PICO ByteDance Ltd.
Computer Vision Algorithm Engineer in PICO MR Team (Full-Time)      • Jun. 2022 - Feb. 2024
Mentor: Xiao Liu
Research Direction: 3D Reconstruction, Monocular Depth Estimation, Stereo Matching and Point Cloud Registration
3D Print Lab
Research Assistant (Intern)      • Jun. 2020 - Aug. 2020
Supervisor: Prof. Hui Zhang
Research Direction: Unity Development