01ABOUT
This is where I build things.
I'm Quan, a first-year Computer Science student in the Advanced Program at the University of Science, VNU-HCM. Look around the room to get to know me.
02NOW
Player stats.
LAST UPDATED · NOV 2026LV.1 · AI STUDENT
Minh-Quan Hoang
CLASS CHANGE: CP → AI
- DOING
- Coding
- AT
- Ho Chi Minh City
- ENERGY
- REPLY IN
- a few hours
Year 1 at APCS, HCMUS
Zalo AI Challenge 2026 LIVE
AI Engineering · Ch. 1
- Started learning AI EngineeringNov 2026
- Joined GDG on Campus HCMUSOct 2026
- Joined the VNOI Contest TeamSep 2026
- Paper accepted · SOICT 2026Sep 2026
- Third Prize · HCMC AI ChallengeSep 2026
03SELECTED WORK
04RESEARCH
CLICK THE PAPER FOR DETAILS
A Multimodal Video Retrieval System with Temporal Anchor Fusion and Temporal Anchor Chaining for Ordered-Event Queries
Search hundreds of hours of video with a description of events in order, like "put an ingredient in the pan, add diced vegetables, then place them in the steamer". The system combines visuals, on-screen text and speech to find the exact frame of each event, in the right order.
@inproceedings{nguyen2026multimodal,
title = {A Multimodal Video Retrieval System with Temporal Anchor Fusion and Temporal Anchor Chaining for Ordered-Event Queries},
author = {Nguyen, Cong-Vinh and Hoang, Minh-Quan and Nguyen, Khai-Hung and Pham-Thi, Thu-Thao and Do, Tien-Minh},
booktitle = {The 15th International Symposium on Information and Communication Technology (SOICT 2026)},
series = {Communications in Computer and Information Science},
publisher = {Springer Nature},
year = {2026}
}A Multimodal Video Retrieval System with Temporal Anchor Fusion and Temporal Anchor Chaining for Ordered-Event Queries
Evidence from every modality is fused at shared time anchors, then events are chained in order with dynamic programming.
First published in SOICT 2026 by Springer Nature.
IN SHORT
Search hundreds of hours of video with a description of events in order. The system combines visuals, on-screen text and speech, then finds the exact frame of each event in the right order.
KEY IDEAS
- Temporal Anchor Fusion merges visual, OCR and speech evidence at shared keyframe anchors
- Temporal Anchor Chaining finds ordered event paths with dynamic programming
- Built for the HCMC AI Challenge 2026: Third Prize, rated Outstanding
MY ROLE
Co-first and corresponding author. I built the data side: keyframes, OCR, speech transcripts and embeddings for 1,487 videos (about 326 hours).
05ACHIEVEMENTS