CV

Doctoral Researcher

The Hong Kong University of Science and Technology (Guangzhou)

Research on audio-visual intelligence, multimodal large language models, model reliability, and assistive speech technologies.

• Developed few-shot audio-visual learning methods using semantic prompts, adapters, latent attention, and prototype regularization.
• Built training-free and semi-training-free cued speech recognition systems with multimodal agents and MLLM-driven hand modeling.
• Studied audio deepfake detection, physically grounded audio-visual generation benchmarks, and evidential uncertainty for in-context learning.
• Contributed to multimodal assessment systems for people with hearing impairment using speech, video, text, and fNIRS signals.

Deep Learning Algorithm Engineer

Suzhou INSNEX Intelligent Technology Co., Ltd.

Developed industrial visual inspection algorithms for defect synthesis and detection.

• Designed GAN-based defect generation methods for data augmentation and industrial inspection; the work contributed to a patent.
• Built generation, detection, and multi-task training pipelines for low-data defect scenarios.

Algorithm Engineer

Shenzhen Research Institute of Big Data

Worked on large-scale multimodal UAV tracking datasets and learning-assisted optimization.

• Co-developed WebUAV-3M, containing 3.3 million frames from 4,500 videos across 223 categories, together with semi-automatic annotation tools and language/audio modalities.
• Explored deep-learning acceleration for mixed-integer linear programming and developed Python evaluation, testing, and visualization tools.

Master's Researcher

Australian National University

Researched unified image style transfer methods.

Research Assistant

CSIRO

Developed interactive 3D geological visualization tools with PyVista and reproducible Binder environments.

Ph.D.

The Hong Kong University of Science and Technology (Guangzhou)

Artificial Intelligence

• Advised by Prof. Li Liu and Prof. Danny H. K. Tsang.
• Research on efficient audio-visual learning, cued speech recognition, audio-visual forgery detection, and language-model uncertainty.

Master

Australian National University

Artificial Intelligence

Bachelor of Engineering

University of Electronic Science and Technology of China

Software Engineering

• Outstanding Graduate.

Semantic Modulated Prompting for Few-Shot Audio-Visual Classification

IEEE/ACM Transactions on Audio, Speech, and Language Processing

Lend a Hand: Semi Training-Free Cued Speech Recognition via MLLM-Driven Hand Modeling

IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Cued-Agent: A Multi-Agent Framework for Automatic Cued Speech Recognition

ACM International Conference on Multimedia

PhyAVBench: A Physical-Centric Audio-Visual Benchmark for Video Generation

ACM International Conference on Multimedia

WebUAV-3M: A Benchmark Unveiling the Power of Million-Scale Deep UAV Tracking

IEEE Transactions on Pattern Analysis and Machine Intelligence

WebUOT-1M: Advancing Deep Underwater Object Tracking with a Million-Scale Benchmark

Advances in Neural Information Processing Systems

Underwater Camouflaged Object Tracking Meets Vision-Language SAM2

CVPR Workshops

Content-Aware Efficient Learner for Audio-Visual Emotion Recognition

International Conference on Social Robotics

4th Place, Efficient Speech Deepfake Detection Challenge 2

IEEE ICME

Student Travel Award

ACM Multimedia

Outstanding Paper Award, CV4Animals Workshop

CV4Animals

Best Student Paper Nomination

ICSR

Outstanding Technology Academic Paper

Shenzhen

Outstanding Graduate

UESTC