Smart Routing for Multimodal Video Retrieval: When to Search What
Fuente:
arXiv
Saved in:
| Main Author: | |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908455371014144 |
|---|---|
| author | Rosa, Kevin Dela |
| author_facet | Rosa, Kevin Dela |
| contents | We introduce ModaRoute, an LLM-based intelligent routing system that dynamically selects optimal modalities for multimodal video retrieval. While dense text captions can achieve 75.9% Recall@5, they require expensive offline processing and miss critical visual information present in 34% of clips with scene text not captured by ASR. By analyzing query intent and predicting information needs, ModaRoute reduces computational overhead by 41% while achieving 60.9% Recall@5. Our approach uses GPT-4.1 to route queries across ASR (speech), OCR (text), and visual indices, averaging 1.78 modalities per query versus exhaustive 3.0 modality search. Evaluation on 1.8M video clips demonstrates that intelligent routing provides a practical solution for scaling multimodal retrieval systems, reducing infrastructure costs while maintaining competitive effectiveness for real-world deployment. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_13374 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Smart Routing for Multimodal Video Retrieval: When to Search What Rosa, Kevin Dela Computer Vision and Pattern Recognition Artificial Intelligence Information Retrieval We introduce ModaRoute, an LLM-based intelligent routing system that dynamically selects optimal modalities for multimodal video retrieval. While dense text captions can achieve 75.9% Recall@5, they require expensive offline processing and miss critical visual information present in 34% of clips with scene text not captured by ASR. By analyzing query intent and predicting information needs, ModaRoute reduces computational overhead by 41% while achieving 60.9% Recall@5. Our approach uses GPT-4.1 to route queries across ASR (speech), OCR (text), and visual indices, averaging 1.78 modalities per query versus exhaustive 3.0 modality search. Evaluation on 1.8M video clips demonstrates that intelligent routing provides a practical solution for scaling multimodal retrieval systems, reducing infrastructure costs while maintaining competitive effectiveness for real-world deployment. |
| title | Smart Routing for Multimodal Video Retrieval: When to Search What |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Information Retrieval |
| url | https://arxiv.org/abs/2507.13374 |