Software Engineering undergraduate at Wuhan University
Actively seeking research collaborations and RA opportunities.
I am interested in LLM reasoning and machine learning, particularly the use of structured representations and reasoning enhancement in computer vision and scientific problem solving. My recent work covers person re-identification, scientific reasoning graphs, and systems that operate under real-world constraints.
Across these projects, I have worked on problem formulation, experimental protocols, failure-case analysis, system implementation, and paper writing. I am actively seeking research collaborations and RA opportunities where I can contribute in these areas.
First author · Accepted to ECCV 2026 · Oct 2025–Mar 2026
This work studies person re-identification when appearance cues become unreliable because of occlusion, clothing changes, or cross-modality. We introduce monocular 3D body geometry as a complementary cue and use a reliability-aware gate to reduce the effect of low-confidence geometry estimates. The method was evaluated through comparisons, ablations, and perturbation tests on nine ReID benchmarks.
First author · WAICA Workshop version accepted · Mar 2026–present
PEARL separates model-generated reasoning graphs from code-based checks for structure, compilation, and source alignment. The preliminary version has been accepted to the WAICA Workshop. The current extension uses curated graphs to train smaller models and is being developed toward an AAAI submission.
First author · In progress, targeting TPAMI · Mar 2026–present
The survey reorganizes appearance variation into short-term perturbations, long-term changes, and compound variations. It also distinguishes appearance robustness from appearance generalization to examine how existing methods and benchmarks cover clothing changes, occlusion, cross-modality, and long-term deployment. Writing, model fine-tuning, and testing are ongoing.
-
Modular Robot System: Responsible for module coordination, bus arbitration, hot-plug discovery, and bus merging. The system reached a 98.5% communication success rate and improved communication efficiency by about 40% over the original design. The project received a National First Prize and placed in the Top 6 at the National University IoT Design Competition.
-
A4R Guided Smartphone AI Assistant: Co-designed a clarify-execute-verify workflow and built more than 150 interactions across six domains and 16 applications. The system achieved a 64.68% completion rate and a 72.33% effective-guidance rate on China Unicom tasks, and the project received a national second prize.
-
SmartPhone Battery: The Moment of Depletion: Combined a physical battery model with ECM-EKF and physics-informed neural-network solvers. Across eight data sources and more than 120,000 time steps, SOC RMSE was 1.24% and overall TTE MAPE was 8.7%. The project received the 2026 MCM Finalist Award.
-
A2A Multi-Agent System: Implemented requirement clarification, protocol adaptation, streaming responses, context propagation, fallback handling, and logging on Holos and A2A v0.3.0 for end-to-end travel-planning and email-collaboration demos.
- Wuhan University, Bachelor's student in Software Engineering, Sep 2023–Jun 2027 (expected)
- GPA: 3.75 / 4.0
- Selected recognition: National First Prize (Top 6) in the National University IoT Design Competition; National Second Prize in the China Youth Sci-Tech Innovation Challenge; Xiaomi Cup National Third Prize; MCM 2026 Finalist; Zheng Geru First-Class Scholarship, ranked first in the grade-level defense; Wuhan University First-Class Scholarship twice
I am actively seeking research collaborations and RA opportunities, especially in computer vision, structured reasoning, and multi-agent systems. Please feel free to contact me at 2023302143001@whu.edu.cn.