Research map / Detection and segmentation / Open-set and few-shot detection
Few-Shot Visual Relationship Co-Localization
Revant Teotia, Vaibhav Mishra, Mayank Maheshwari, Anand Mishra
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 |
|---|---|---|---|
| add Relation Network similarity instead of cosine similarity (ablation) | VtransE embedding · VrR-VG | VR-CorLoc: 73.34 -> 78.99 (bag 2); 74.20 -> 76.12 (bag 4); 82.56 -> 84.07 (bag 8) | Improved |
| add VTransE embedding instead of concatenation (ablation) | Relation Network model · VrR-VG | VR-CorLoc: 75.61 -> 78.99 (bag 2); 74.02 -> 76.12 (bag 4); 76.38 -> 84.07 (bag 8) | Improved |
| add both VTransE and Relation Network over concat + cosine (ablation) | few-shot VRC framework · VrR-VG | VR-CorLoc: 72.16 -> 78.99 (bag 2); 70.86 -> 76.12 (bag 4); 76.85 -> 84.07 (bag 8) | Improved |
| add Relation Network instead of cosine similarity (ablation) | VtransE embedding · VrR-VG | Bag-CorLoc: 59.84 -> 63.40 (bag 2); 36.23 -> 48.06 (bag 4); 34.64 -> 45.48 (bag 8) | Improved |
| add VTransE embedding instead of concatenation (ablation) | Relation Network model · VG-150 | Bag-CorLoc: 54.55 -> 61.10 (bag 2) | Improved |
| add Relation Network instead of cosine similarity (ablation) | VtransE embedding · VG-150 | Bag-CorLoc: 55.67 -> 61.10 (bag 2) | Improved |
| give subject bounding box in all bag images (ablation) | Our best model · VrR-VG | VR-CorLoc: 78.99 -> 83.90 (bag 2); 76.12 -> 88.25 (bag 4) | Improved |
| give subject and object box in one bag image (ablation) | Our best model · VrR-VG | VR-CorLoc: 78.99 -> 87.44 (bag 2); 76.12 -> 84.46 (bag 4) | Improved |
| increase bag size from 2 to 8 (ablation) | Our best model · VrR-VG | Bag-CorLoc: 63.40 -> 48.06 -> 45.48 | Hurt |
| increase bag size from 2 to 8 (ablation) | Our best model · VrR-VG | VR-CorLoc: 78.99 -> 76.12 -> 84.07 | Mixed |
Closest papers by meaning
- Few-Shot Referring Relationships in Videos (CVPR 2023 · similarity 0.90)
- Localize, Assemble, and Predicate: Contextual Object Proposal Embedding for Visual Relation Detection (AAAI 2020 · similarity 0.89)
- VrR-VG: Refocusing Visually-Relevant Relationships (ICCV 2019 · similarity 0.88)
- Tensorize, Factorize and Regularize: Robust Visual Relationship Learning (CVPR 2018 · similarity 0.87)
- Visual Relationship Detection with Language Priors (ECCV 2016 · similarity 0.87)
- Zoom-Net: Mining Deep Feature Interactions for Visual Relationship Recognition (ECCV 2018 · 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.
