Research map / Detection and segmentation / Open-set and few-shot detection
FindIt: Generalized Localization with Natural Language Queries
Weicheng Kuo, Fred Bertsch, Wei Li, AJ Piergiovanni, Mohammad Saffar, Anelia Angelova
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 multi-task training (REC+LOC+DET) vs single-task REC | FindIt · RefCOCO val | precision@1: 79.45 -> 84.66 | Improved |
| train on all RefCOCO splits plus LOC and DET (FindIt-MIX) | FindIt · COCO text-based localization | AP50: 77.9 -> 78.6 | Improved |
| add multi-task training to detection | FindIt · COCO detection | mAP: 40.6 -> 39.7 | Hurt |
| multi-task training vs single-task LOC on novel categories | FindIt-MIX · Objects365 Novel-285 | AP50: 15.2 -> 18.7 | Improved |
| multi-task training vs single-task LOC on super-categories | FindIt-MIX · Objects365 super-categories | AP50: 15.3 -> 17.5 | Improved |
| use attention fusion rather than concat fusion (ablation) | FindIt · RefCOCO-UNC | precision@1: 76.8 -> 78.6 | Improved |
| use attention fusion rather than product fusion (ablation) | FindIt · RefCOCO-UNC | precision@1: 76.0 -> 78.6 | Improved |
| add second fusion level (4,5) instead of only level 5 (1024 dim, 12 layers) (ablation) | FindIt · REC (ablation) | precision@1: 80.5 -> 82.1 | Improved |
| add second fusion level (4,5) instead of only level 5 (1024 dim, 12 layers) (ablation) | FindIt · Detection (ablation) | mAP: 33.0 -> 35.7 | Improved |
| scale attention fusion from 256 dim/3 layers to 1024 dim/12 layers (ablation) | FindIt · REC (ablation) | precision@1: 79.3 -> 82.1 | Improved |
| scale attention fusion from 256 dim/3 layers to 1024 dim/12 layers (ablation) | FindIt · Detection (ablation) | mAP: 35.6 -> 35.7 | No clear effect |
| change LOC prompt from 'Find the X' to 'X' (ablation) | FindIt · COCO LOC (ablation) | AP50: 78.78 -> 79.13 | No clear effect |
Closest papers by meaning
- Mechanisms of Object Localization in Vision-Language Models (CVPR 2026 · similarity 0.88)
- RELOCATE: A Simple Training-Free Baseline for Visual Query Localization Using Region-Based Representations (CVPR 2025 · similarity 0.88)
- Grounding Everything: Emerging Localization Properties in Vision-Language Transformers (CVPR 2024 · similarity 0.88)
- Text to Point Cloud Localization with Multi-Level Negative Contrastive Learning (AAAI 2025 · similarity 0.88)
- Recognize Any Regions (NeurIPS 2024 · similarity 0.87)
- Your Large Vision-Language Model Only Needs A Few Attention Heads For Visual Grounding (CVPR 2025 · 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.
