Research map / Detection and segmentation / Open-set and few-shot detection
Zero-Shot Object Counting
Jingyi Xu, Hieu Le, Vu Nguyen, Viresh Ranjan, Dimitris Samaras
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 class-name-guided patch selection instead of RPN exemplars | Exemplar-free counting on FSC-147 · FSC-147 test | RMSE: 14.52 | Improved |
| Add class-prototype filtering to random patch selection (ablation) | Ours (Base) · FSC-147 val | MAE: 35.20 -> 28.01 | Improved |
| Add class-prototype filtering to random patch selection (ablation) | Ours (Base) · FSC-147 test | MAE: 31.37 -> 25.30 | Improved |
| Add error-predictor exemplar selection to random patch selection (ablation) | Ours (Base) · FSC-147 val | MAE: 35.20 -> 27.98 | Improved |
| Add error-predictor exemplar selection to random patch selection (ablation) | Ours (Base) · FSC-147 test | MAE: 31.37 -> 23.80 | Improved |
| Combine class-prototype filtering with error-predictor selection (ablation) | Ours (Base) · FSC-147 val | MAE: 35.20 -> 26.93 | Improved |
| Combine class-prototype filtering with error-predictor selection (ablation) | Ours (Base) · FSC-147 test | MAE: 31.37 -> 22.09 | Improved |
| Use prototype-selected patches instead of random patches as exemplars | FamNet · FSC-147 val | MAE: 5.18 | Improved |
| Use prototype-selected patches instead of random patches as exemplars | BMNet · FSC-147 val | MAE: 8.59 | Improved |
| Use prototype-selected patches instead of random patches as exemplars | BMNet+ · FSC-147 val | MAE: 5.60 | Improved |
| Add error-predictor selection after prototype filtering | FamNet · FSC-147 val | MAE: 1.76 | Improved |
| Add error-predictor selection after prototype filtering | BMNet · FSC-147 val | MAE: 1.00 | Improved |
Closest papers by meaning
- Point Segment and Count: A Generalized Framework for Object Counting (CVPR 2024 · similarity 0.88)
- Referring Expression Counting (CVPR 2024 · similarity 0.87)
- PBECount: Prompt-Before-Extract Paradigm for Class-Agnostic Counting (AAAI 2025 · similarity 0.87)
- VLCounter: Text-Aware Visual Representation for Zero-Shot Object Counting (AAAI 2024 · similarity 0.87)
- Vision Transformer Off-the-Shelf: A Surprising Baseline for Few-Shot Class-Agnostic Counting (AAAI 2024 · similarity 0.87)
- A Fixed-Point Approach to Unified Prompt-Based Counting (AAAI 2024 · similarity 0.86)
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.
