Research map / Detection and segmentation / Open-set and few-shot detection
Sketch-Guided Object Localization in Natural Images
Aditay Tripathi, Rajath R. Dani, Anand Mishra, Anirban Chakraborty
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 |
|---|---|---|---|
| use cross-modal attention instead of Matchnet's non-local plus co-excitation (ablation) | one-shot sketch localization · COCO val2017 common | AP@50 (%): 48.5 -> 50.0 (mAP 0.28 -> 0.30) | Improved |
| use cross-modal attention instead of modified Faster R-CNN (ablation) | one-shot sketch localization · COCO val2017 common | AP@50 (%): 31.5 -> 50.0 | Improved |
| use cross-modal attention vs modified Faster R-CNN on small VOC data (ablation) | one-shot sketch localization · VOC test2007 | AP@50 (%): 0.65 -> 0.65 | No clear effect |
| use cross-modal attention instead of Matchnet (ablation) | one-shot sketch localization · VOC test2007 | AP@50 (%): 0.61 -> 0.65 | Improved |
| use cross-modal attention instead of Matchnet on unseen classes (ablation) | one-shot sketch localization · COCO disjoint unseen | AP@50 (%): 12.4 -> 15.0 | Improved |
| use cross-modal attention instead of Matchnet on seen classes (ablation) | one-shot sketch localization · COCO disjoint seen | AP@50 (%): 49.1 -> 48.8 | No clear effect |
| add query fusion with 5 sketch queries (ablation) | cross-modal attention · COCO common | AP@50 (%): 50.0 -> 52.6 (mAP 0.30 -> 0.32) | Improved |
| add attention fusion with 5 sketch queries (ablation) | cross-modal attention · COCO common | AP@50 (%): 50.0 -> 53.1 (mAP 0.30 -> 0.32) | Improved |
| add attention fusion with 5 queries on unseen classes (ablation) | cross-modal attention · COCO disjoint unseen | AP@50 (%): 15.0 -> 17.1 | Improved |
| add attention fusion with 3 queries (ablation) | cross-modal attention · COCO common | AP@50 (%): 50.0 -> 52.0 (mAP 0.30 -> 0.31) | Improved |
| add margin-ranking loss to classification loss (ablation) | cross-modal attention · COCO common | AP@50 (%): 40.0 -> 50.0 | Improved |
| add classification loss to margin-ranking loss (ablation) | cross-modal attention · COCO common | AP@50 (%): 41.4 -> 50.0 | Improved |
Closest papers by meaning
- Where Is My Wallet? Modeling Object Proposal Sets for Egocentric Visual Query Localization (CVPR 2023 · similarity 0.88)
- What Can Human Sketches Do for Object Detection? (CVPR 2023 · similarity 0.88)
- Investigating the Role of Image Retrieval for Visual Localization (IJCV 2022 · similarity 0.87)
- FindIt: Generalized Localization with Natural Language Queries (ECCV 2022 · similarity 0.87)
- RELOCATE: A Simple Training-Free Baseline for Visual Query Localization Using Region-Based Representations (CVPR 2025 · similarity 0.86)
- Composite Sketch+Text Queries for Retrieving Objects with Elusive Names and Complex Interactions (AAAI 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.
