VideoEase at VBS2025: An Interactive Video Retrieval System
Lists
Only you see this
Abstract
We present the VideoEase interactive video retrieval system, which we used to participate in VBS2025. This is the first time that VideoEase has taken part in the VBS challenge. VideoEase is built on the Milvus vector database, which supports vector searches on massive datasets with high-dimensional vectors. The CLIP, BLIP2, and OpenCLIP models play a crucial role in encoding keyframe images and queries into embeddings. A new user interface with simple yet effective components is also introduced. We experiment with VBS24 queries to evaluate the performance of the VideoEase system. Our results show that answer rank is improved by merging outputs from several models with appropriate weights.
BibTeX
@inproceedings{tran2025videoease,
author = {Quang-Linh Tran and Binh T. Nguyen and Gareth J. F. Jones and Cathal Gurrin},
title = {{VideoEase at VBS2025: An Interactive Video Retrieval System}},
booktitle = {MMM},
year = {2025},
doi = {10.1007/978-981-96-2074-6\_44},
}| Fusionista2.0: Efficiency Retrieval System for Large-Scale Datasets | Lecture Notes in Computer Science · 2026 |
| AI-based Video Content Understanding for Automatic and Interactive Multimedia Retrieval | Workshops @ CVPR · 2025 |
| What Matters for Ad-hoc Video Search? A Large-scale Evaluation on TRECVID | arXiv · 2021 |
| Towards Efficient and Robust Moment Retrieval System: A Unified Framework for Multi-Granularity Models and Temporal Reranking | Workshops @ CVPR · 2025 |
| RUC_AIM3 at TRECVID 2024: Ad-hoc Video Search | Trec · 2024 |
| Can Relevance Feedback, Conversational Search and Foundation Models Work Together for Interactive Video Search and Exploration? | Workshops @ CVPR · 2025 |