Abstract
The explosive proliferation of online video content necessitates advanced retrieval systems, as promoted by the Ho Chi Minh AI Challenge (AIC) 2025. This competition comprises three tasks: Known-Item Search (KIS), Visual Question Answering (VQA), and the newly introduced Temporal Retrieval and Alignment of Key Events (TRAKE).
To address these challenges, we propose a comprehensive multimodal retrieval framework. Our system integrates heterogeneous data sources, including semantic embeddings from CLIP and BEiT-3, Optical Character Recognition (OCR), Automatic Speech Recognition (ASR), and open-vocabulary object detection. The architecture features a dual-layer search mechanism. For discrete queries (KIS, VQA), a Multi-modal Search layer retrieves and ranks results from parallel data streams using the Weighted Reciprocal Rank Fusion (WRRF) algorithm. To address the sequential nature of the TRAKE task, we introduce a novel Temporal Search module built upon an efficient K-pointer sequential re-ranking algorithm. This algorithm effectively validates and ranks video segments containing events in a specified temporal order. Finally, we present a competition-oriented user interface designed for real-time interaction, supporting multi-stage query construction and precise temporal refinement tools.
Authors
Khoa Dinh Duc Anh, Duc-Tai Dinh, Trung Nguyen Le Hoang, Nhan Nguyen Thanh — Team WATLERE, University of Science, VNU-HCM.
Highlights
- Best Presentation Award at SOICT 2025 (Session 10A: Lifelog Event Retrieval).
- Top 12+/80+ teams at AI Challenge HCMC 2025.
- Novel K-pointer sequential re-ranking algorithm for the TRAKE temporal retrieval task.
- Weighted Reciprocal Rank Fusion (WRRF) for multi-stream fusion across CLIP/BEiT-3 embeddings, OCR, ASR, and open-vocabulary detection.
- Competition-oriented interactive UI for multi-stage query construction and temporal refinement.
References
[Springer Nature] KPTER: K-Pointer for Temporal Event Retrieval
SOICT 2025 — The 14th International Symposium on Information and Communication Technology. Presented Saturday, December 13th, 2025.
BibTeX
@inproceedings{kpter2025, title = {KPTER: K-Pointer for Temporal Event Retrieval}, author = {Dinh, Khoa Duc Anh and Dinh, Duc-Tai and Nguyen, Trung Le Hoang and Nguyen, Nhan Thanh}, booktitle = {Proceedings of the 14th International Symposium on Information and Communication Technology (SOICT)}, year = {2025}, note = {Best Presentation Award}}
