Research map / Detection and segmentation / Weakly supervised detection and segmentation
Discriminative Triad Matching and Reconstruction for Weakly Referring Expression Grounding
Mingjie Sun, Jimin Xiao, Eng Gee Lim, Si Liu, John Yannis Goulermas
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 target-unit reconstruction loss (ablation) | Discriminative triad WREG network · RefCOCO (cross-eval) | accuracy: 29.5 -> 43.2 | Improved |
| add target-unit reconstruction loss (ablation) | Discriminative triad WREG network · RefCOCOg (cross-eval) | accuracy: 27.0 -> 39.8 | Improved |
| add discriminative-unit loss (ablation) | Discriminative triad WREG network · RefCOCO (cross-eval) | accuracy: 38.1 -> 43.2 | Improved |
| add reference-unit loss (ablation) | Discriminative triad WREG network · RefCOCO (cross-eval) | accuracy: 37.5 -> 43.2 | Improved |
| add triad-level reconstruction module (ablation) | Discriminative triad WREG network · RefCOCO (cross-eval) | accuracy: 35.6 -> 43.2 | Improved |
| use Gumbel-softmax hard reconstruction instead of soft (ablation) | Discriminative triad WREG network · RefCOCO (cross-eval) | accuracy: 42.1 -> 43.2 | Improved |
| use Gumbel-softmax hard reconstruction instead of soft (ablation) | Discriminative triad WREG network · RefCOCO+ (cross-eval) | accuracy: 39.2 -> 40.1 | Improved |
| use all triads at inference instead of a single random triad (ablation) | Discriminative triad WREG network · RefCOCO (cross-eval) | accuracy: 39.6 -> 43.2 | Improved |
| use all triads at inference instead of a single random triad (ablation) | Discriminative triad WREG network · RefCOCOg (cross-eval) | accuracy: 36.1 -> 39.8 | Improved |
| use detected proposals instead of ground-truth boxes | Discriminative triad WREG network · RefCOCO val | accuracy: 39.21 -> 38.35 | Hurt |
| replace KPRN with triad-level matching and reconstruction | Discriminative triad WREG network · RefCOCOg val | accuracy: 35.44 -> 43.24 | Improved |
Closest papers by meaning
- RefDetector: A Simple Yet Effective Matching-based Method for Referring Expression Comprehension (AAAI 2025 · similarity 0.88)
- Referencing Where to Focus: Improving Visual Grounding with Referential Query (NeurIPS 2024 · similarity 0.88)
- Rethinking Two-Stage Referring Expression Comprehension: A Novel Grounding and Segmentation Method Modulated by Point (AAAI 2024 · similarity 0.88)
- Boosting Weakly Supervised Referring Image Segmentation via Progressive Comprehension (NeurIPS 2024 · similarity 0.88)
- Hierarchical Alignment-enhanced Adaptive Grounding Network for Generalized Referring Expression Comprehension (AAAI 2025 · similarity 0.87)
- Mask Grounding for Referring Image Segmentation (CVPR 2024 · 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.
