Research map / Vision-language and multimodal / Cross-modal learning
Adaptive Cross-Modal Embeddings for Image-Text Alignment
Jonatas Wehrmann, Camila Kolling, Rodrigo C Barros
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 |
|---|---|---|---|
| ADAPT-T2I single model vs SCAN | image-text alignment · Flickr30k test | Image Retrieval R@1: 45.8 -> 57.0 | Improved |
| ADAPT-T2I vs SCAN-ens | image-text alignment · MS COCO 1k test | Image Retrieval R@1: 58.8 -> 63.3 | Improved |
| add fovea module (ablation) | ADAPT-T2I · Flickr30k val | Rsum: 478.2 -> 494.2 | Improved |
| add fovea module (ablation) | ADAPT-I2T · Flickr30k val | Rsum: 427.4 -> 482.0 | Improved |
| increase fovea lambda from 1 to 10 (ablation) | ADAPT-T2I · Flickr30k val | Rsum: 489.8 -> 494.2 | Improved |
| learn lambda by backprop instead of fixed 10 (ablation) | ADAPT-T2I · Flickr30k val | Rsum: 494.2 -> 493.1 | No clear effect |
| add fixed GloVe to random trainable embeddings (ablation) | ADAPT-T2I · Flickr30k val | Rsum: 484.1 -> 494.2 | Improved |
| fine-tune GloVe instead of keeping fixed (ablation) | ADAPT-T2I · Flickr30k val | Rsum: 494.2 -> 492.3 | No clear effect |
| text-guided image adaptation (T2I) instead of image-guided text (I2T) | ADAPT · Flickr30k test | Rsum: 484.8 -> 494.8 | Improved |
| ADAPT vs SCAN on retrieval time | image-text alignment · Flickr30k | speed: same R@1 but 25x faster | Improved |
Closest papers by meaning
- Improving Cross-Modal Alignment with Synthetic Pairs for Text-Only Image Captioning (AAAI 2024 · similarity 0.85)
- Cross-Modal and Uni-Modal Soft-Label Alignment for Image-Text Retrieval (AAAI 2024 · similarity 0.85)
- Aligning Text to Image in Diffusion Models is Easier Than You Think (NeurIPS 2025 · similarity 0.84)
- Linguistic-Aware Patch Slimming Framework for Fine-grained Cross-Modal Alignment (CVPR 2024 · similarity 0.84)
- Multi-Level Cross-Modal Alignment for Image Clustering (AAAI 2024 · similarity 0.84)
- Seeing the Image: Prioritizing Visual Correlation by Contrastive Alignment (NeurIPS 2024 · similarity 0.83)
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.
