Research map / Vision-language and multimodal / Human and embodied modeling
Instance-Aware Image and Sentence Matching With Selective Multimodal LSTM
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 selective instance attention instead of mean-vector representations (ablation) | sm-LSTM · Flickr30K | Sum: 261.5 -> 343.4 | Helped |
| Use selective instance attention instead of mean-vector representations (ablation) | sm-LSTM · Microsoft COCO | Sum: 331.9 -> 421.5 | Helped |
| Add global context to the attention scheme (ablation) | sm-LSTM · Flickr30K | Sum: 271.1 -> 343.4 | Helped |
| Add global context to the attention scheme (ablation) | sm-LSTM · Microsoft COCO | Sum: 359.5 -> 421.5 | Helped |
| Add the attention scheme while retaining global context (ablation) | sm-LSTM · Flickr30K | Sum: 300.1 -> 343.4 | Helped |
| Add the attention scheme while retaining global context (ablation) | sm-LSTM · Microsoft COCO | Sum: 376.0 -> 421.5 | Helped |
| Ensemble the four sm-LSTM variants (ablation) | sm-LSTM · Flickr30K | Sum: 343.4 -> 358.7 | Helped |
| Ensemble the four sm-LSTM variants (ablation) | sm-LSTM · Microsoft COCO | Sum: 421.5 -> 431.8 | Helped |
| Use 3 timesteps instead of 1 (ablation) | sm-LSTM · Flickr30K | image annotation R@1: 38.8 -> 42.4 | Helped |
| Increase timesteps from 3 to 4 (ablation) | sm-LSTM · Flickr30K | image annotation R@1: 42.4 -> 38.2 | Hurt |
| Increase timesteps from 3 to 5 (ablation) | sm-LSTM · Flickr30K | image annotation R@1: 42.4 -> 38.1 | Hurt |
| Add pairwise doubly stochastic regularization (ablation) | sm-LSTM · Flickr30K | image annotation R@1: 37.9 -> 42.4 | Helped |
Closest papers by meaning
- Multi-Modality Cross Attention Network for Image and Sentence Matching (CVPR 2020 · similarity 0.89)
- Expressing Objects Just Like Words: Recurrent Visual Embedding for Image-Text Matching (AAAI 2020 · similarity 0.88)
- IMRAM: Iterative Matching With Recurrent Attention Memory for Cross-Modal Image-Text Retrieval (CVPR 2020 · similarity 0.88)
- Learning Semantic Concepts and Order for Image and Sentence Matching (CVPR 2018 · similarity 0.88)
- GOAL: Global-local Object Alignment Learning (CVPR 2025 · similarity 0.86)
- Towards General Continuous Memory for Vision-Language Models (NeurIPS 2025 · similarity 0.85)
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.