Look Beyond Saliency: Low-Attention Guided Dual Encoding for Video Semantic Search
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866917468257124352 |
|---|---|
| author | Aljehrai, Faisal Alkhrashi, Mohammed A. Almuhrij, Alreem Abuhimed, Sarah Aldossary, Noorh Aldwyish, Abdullah Aljadaany, Raied Alamri, Huda Khan, Muhammad Kamran J |
| author_facet | Aljehrai, Faisal Alkhrashi, Mohammed A. Almuhrij, Alreem Abuhimed, Sarah Aldossary, Noorh Aldwyish, Abdullah Aljadaany, Raied Alamri, Huda Khan, Muhammad Kamran J |
| contents | Video semantic search in densely crowded scenes remains a challenging task due to visual encoders tendency to prioritize salient foreground regions while neglecting contextually important, background areas. We propose an Inverse Attention Embedding mechanism that explicitly captures and highlights these overlooked regions. By combining inverse attention embeddings with traditional visual embeddings, our method significantly enhances semantic retrieval performance without additional training. Initial experiments and ablation studies demonstrate promising improvements over existing approaches in recall for video semantic search in crowded environments. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_06229 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Look Beyond Saliency: Low-Attention Guided Dual Encoding for Video Semantic Search Aljehrai, Faisal Alkhrashi, Mohammed A. Almuhrij, Alreem Abuhimed, Sarah Aldossary, Noorh Aldwyish, Abdullah Aljadaany, Raied Alamri, Huda Khan, Muhammad Kamran J Computer Vision and Pattern Recognition Video semantic search in densely crowded scenes remains a challenging task due to visual encoders tendency to prioritize salient foreground regions while neglecting contextually important, background areas. We propose an Inverse Attention Embedding mechanism that explicitly captures and highlights these overlooked regions. By combining inverse attention embeddings with traditional visual embeddings, our method significantly enhances semantic retrieval performance without additional training. Initial experiments and ablation studies demonstrate promising improvements over existing approaches in recall for video semantic search in crowded environments. |
| title | Look Beyond Saliency: Low-Attention Guided Dual Encoding for Video Semantic Search |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2605.06229 |