Research map / Detection and segmentation / Open-set and few-shot detection
On Exploring Undetermined Relationships for Visual Relationship Detection
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 |
|---|---|---|---|
| Add generated undetermined relationships to training (ablation) | MFLN on VRD · VRD relation detection, entire set | R50 recall: 17.36 -> 23.89 | Helped |
| Replace undetermined-relationship training with triplet NMS (ablation) | MFLN on VRD · VRD relation detection, entire set | R50 recall: 17.36 -> 15.53 | Hurt |
| Use pair filtering instead of generated undetermined relationships (ablation) | MFLN on VRD · VRD relation detection, entire set | R50 recall: 17.36 -> 21.58 | Helped |
| Add visual, spatial, and linguistic modalities rather than using only one modality (ablation) | MF-URLN on VRD · VRD relation detection | R50 recall: 22.64; 18.49; 17.94 -> 23.89 | Mixed |
| Transform features to equal dimensions before concatenating rather than directly concatenate (ablation) | MF-URLN on VRD · VRD predicate and relation detection | R50 recall: 58.22 vs 55.77; 23.89 vs 22.61 | Mixed |
| Remove determinate-confidence information from the relationship detection subnetwork (ablation) | MF-URLN on VRD · VRD relation detection | R50 recall | Hurt |
| Add an inference model that combines union-, subject-, and object-level relationship predictions (ablation) | MF-URLN on the VRD zero-shot set · VRD zero-shot predicate detection | R50/100 recall: 26.9 -> 27.2 | Helped |
| Add an inference model that combines union-, subject-, and object-level relationship predictions (ablation) | MF-URLN on the VRD zero-shot set · VRD zero-shot phrase detection | R100 recall: 7.9 -> 9.2 | Helped |
| Add an inference model that combines union-, subject-, and object-level relationship predictions (ablation) | MF-URLN on the VRD zero-shot set · VRD zero-shot relation detection | R100 recall: 5.5 -> 6.4 | Helped |
| Use generated undetermined relationships | MF-URLN on VRD · VRD phrase detection | R50 recall: 25.2 -> 31.5 | Helped |
| Use generated undetermined relationships | MF-URLN on VRD · VRD relation detection | R50 recall: 19.5 -> 23.9 | Helped |
Closest papers by meaning
- Visual Relationship Detection With Deep Structural Ranking (AAAI 2018 · similarity 0.91)
- Large-Scale Visual Relationship Understanding (AAAI 2019 · similarity 0.90)
- Tensorize, Factorize and Regularize: Robust Visual Relationship Learning (CVPR 2018 · similarity 0.90)
- Multi-Modal Prompting for Open-Vocabulary Video Visual Relationship Detection (AAAI 2024 · similarity 0.87)
- Groupwise Query Specialization and Quality-Aware Multi-Assignment for Transformer-based Visual Relationship Detection (CVPR 2024 · similarity 0.87)
- Towards Flexible Visual Relationship Segmentation (NeurIPS 2024 · 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.