Porthos Gaa-Lok Fu

Academic Brief of Porthos Gaa-Lok Fu

Greetings! This page is the academic brief of Porthos Gaa-Lok Fu. Click for Chinese version here 点击此处查看中文版本

Porthos Gaa-Lok Fu is an undergraduate student from Central South University, currently majored in Mechanical Engineering with Transportation. His research interest is in deep learning, computer vision, and solving VQA tasks.

He is also intersted in 3D printing, mechanical manufacturing.

1. Education Background

1.1. Central South University 2022 - present (expected: 2026)

1.1.1. Degree (expected): Bachelor of Engineering (Mechanical Engineering with Transportation)

1.1.2. Ranking: 2/29, with avg. Score: 92.09/100 (till 16th May, 2025)

1.1.3. Awards:

National Scholarship, for school year of 22/23
First-Class Scholarship of CSU, for school year of 22/23
Outstanding Student Award, for school year of 22/23
Corporate Scholarship (Tongguan 铜冠, by Tongling Nonferrous Metals Group), for school year of 23/24
Second-Class Scholarship of CSU, for school year of 23/24

1.2. Hangzhou Xuejun High School (Xixi Campus) 2019 - 2022

OI experience, Second Prize for CSP-S 2019 and 2020, Second Prize for NOIP 2020, all in Zhejiang Province. See also: Introduction from CCF-NOI.

2. Research Experience

Mentor: Dr Kan, School of Computer Science, CSU

Start: September 2023

Weekly Commitment: Over 15 hours

Focus Areas: Models and methods in multimodal large language models and their applications in downstream tasks.

3. Projects

3.1. High-Resolution Remote Sensing Image VQA

Utilizing the A* algorithm to analyze weights for VQA. This project assesses the relevance between questions, options, and image slices via heatmap generator or vision grounding models, ultimately selecting the optimal slice for answering. (Preliminary completed, pending refinement and submission to ACM Multimedia, First Author or Student First Author).

3.2. “Object Retrieval for VQA with Outside Knowledge” (Third Author)

Proposed an object retrieval task for visual question answering using external knowledge, integrating LLM for enhanced image content understanding. Responsible for part of the training and the entire benchmark design. Submitted to Pattern Recognition.

3.3. Traffic Safety Assessment Project

Evaluating non-motor vehicle road safety (e.g., passenger count, helmet usage) using deep learning methods and live camera feeds, in collaboration with Felix Wilhelm Siebert’s team at the Denmark Technical University. Currently in the data downloading and processing phase, with future training tasks planned.

3.4. Traffic Technology Competition: Rail Track Status Detector

First prize of School Competition, Team Leader

4. Competitions (Selected)

Central South University Programming Competitions, first prize in 16th, 17th and 18th competition (2023, 2024 and 2025, respectively)
Group Programming Ladder Competition, third prize as individual (2024)
Commercial Elite Challenge: Innovation and Entrepreneurship Competition, 1st prize (2024)
Asia and Pacific Mathematical Contest in Modeling, third prize (2022)

5. English Proficiency

Passed CET-4 (600) and CET-6 (523) in the first year. Preparing IELTS.

Fluent in spoken English, proficient in reading academic papers, and possess a certain level of academic writing skills.

6. Programming Skills

Accumulated over 2000 hours of C++ programming experience during high school and university.

Proficient in utilizing models and code from Hugging Face, familiar with tools such as Conda, Google Colab, and PyTorch.

Capable of quickly reproducing others’ work and modifying.

7. Additional Information

7.1. Self-Evaluation

Cheerful personality with leadership skills and a strong spirit of teamwork. High learning ability and a strong exploratory mindset towards new concepts.

7.2. Extracurricular

Served as the Vice Minister of the Publicity Department in the Student Union, for school year of 23/24

7.3. Leadership Roles:

Frequently appointed as Coordinator in team projects and exams, leading teams to efficiently complete tasks and achieve excellent results.