Research map / Vision-language and multimodal / Human and embodied modeling
HAL: Improved Text-Image Matching by Mitigating Visual Semantic Hubs
Fangyu Liu, Rongtian Ye, Xun Wang, Shuaipeng Li
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use HAL instead of Max | SCAN · MS-COCO | R@1, image→text: 70.9 -> 78.3 | Improved |
| Use HAL instead of Max | SCAN · MS-COCO | rsum: 500.5 -> 512.7 | Improved |
| Use HAL instead of Sum | GRU+VGG19 · Flickr30k | rsum: 291.0 -> 320.0 | Improved |
| Use HAL instead of Max | GRU+VGG19 · Flickr30k | rsum: 281.4 -> 320.0 | Improved |
| Add global memory-bank weighting to HAL (ablation) | GRU+VGG19 · MS-COCO c2 (5k) | rsum: 436.1 -> 439.5 | Improved |
| Add global memory-bank weighting to HAL (ablation) | GRU+IRv2 · MS-COCO c2 (5k) | rsum: 453.2 -> 458.5 | Improved |
| Add global memory-bank weighting to HAL (ablation) | GRU+ResNet152 · MS-COCO c2 (5k) | rsum: 462.0 -> 466.7 | Improved |
| Add global memory-bank weighting to HAL (ablation) | GRU+ResNet152 · MS-COCO c1 (1k) | rsum: 469.2 -> 476.6 | Improved |
| Increase batch size to expand the local neighborhood (ablation) | HAL · MS-COCO | rsum | Improved |
| Use a memory bank sized at 5% of training data (ablation) | HAL + memory bank · MS-COCO | rsum | Improved |
Closest papers by meaning
- Consensus-Aware Visual-Semantic Embedding for Image-Text Matching (ECCV 2020 · similarity 0.91)
- Composing Object Relations and Attributes for Image-Text Matching (CVPR 2024 · similarity 0.89)
- Image-to-Image Matching via Foundation Models: A New Perspective for Open-Vocabulary Semantic Segmentation (CVPR 2024 · similarity 0.88)
- Asymmetric Visual Semantic Embedding Framework for Efficient Vision-Language Alignment (AAAI 2025 · similarity 0.88)
- SD4Match: Learning to Prompt Stable Diffusion Model for Semantic Matching (CVPR 2024 · similarity 0.87)
- Expanding the Scope of Negatives: Boosting Image-Text Matching with Negatives Distribution Guided Learning (AAAI 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
