Zitong Qi齐梓桐
Mathematics undergraduate at Ningbo University
I am an undergraduate in Mathematics and Applied Mathematics at Ningbo University. My current work covers video understanding in real-world systems, multimodal representation learning, and lightweight vision-language navigation at EAST Lab, Eastern Institute of Technology, Ningbo.
SafeCommunity AI and the ACVF multimodal medical-imaging paper.
A multi-channel intelligent video analytics system spanning tracking, rule-based alerts, anomaly recognition, and vision-language review.
Led the core algorithms and system integration, connecting detection and tracking, ROI rules, video anomaly recognition, and vision-language review in one web workflow.
PyTorch · VideoMAE · CLIP · Qwen-VL · QLoRA · Flask · ByteTrack
IEEE BIBM 2026
Aligning frozen 3D brain MRI representations with AAL regional volumes and clinical information for lightweight multimodal enhancement and leakage-aware evaluation.
Multimodal method design, experiments, and evaluation
3D MRI · Contrastive learning · BioClinicalBERT · Leakage-aware evaluation · PyTorch
Ningbo University, Yongjiang Laboratory, and Eastern Institute of Technology, Ningbo.
B.Sc. in Mathematics and Applied Mathematics
Building a mathematical foundation while working on machine learning, computer vision, and multimodal methods.
SafeCommunity AI
Core algorithms and system engineering
Integrated multi-stream video, detection and tracking, rule-based alerts, video anomaly recognition, and VLM review into one system.
EAST Lab · Wei Zhang Group
Research Assistant
Working on vision-language navigation (VLN), including evaluation reproduction, model-behavior analysis, and lightweight embodied-model experiments.
Methods and tools evidenced by project work and local repositories.
Python · TypeScript / JavaScript · HTML / CSS · Bash / PowerShell
PyTorch · Transformers · PEFT / QLoRA · CLIP · Qwen-VL
OpenCV · YOLOv8 · VideoMAE v2 · SAM 2 · ByteTrack
Habitat-Lab · Habitat-Sim · VLN-CE · Isaac Lab · Isaac Sim
PPO / Actor-Critic · Hierarchical policies · Multi-objective rewards · LSTM / Transformer Policy · Reward Shaping
NumPy · Pandas · SciPy · scikit-learn · XGBoost
Competition awards and scholarship.
Led hierarchical PPO, reward design, and training iteration.
Contributed to solution development and the engineering presentation.
Worked on problem formulation, numerical analysis, and technical writing.
Awarded for coursework and overall performance.
Strength training, long routes, and collaborative puzzles.
Regular training helps me keep a steady long-term rhythm.
I enjoy getting to know cities and landscapes through long routes.
I enjoy building hypotheses from limited evidence and reasoning together.
For research conversations or internships, reach out on GitHub.