Research map / Detection and segmentation / Weakly supervised detection and segmentation
Counting With Focus for Free
Zenglin Shi, Pascal Mettes, Cees G. M. Snoek
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 focus from segmentation (ablation) | base counting network · ShanghaiTech Part_A | MAE: 74.8 -> 72.3 | Improved |
| add spatial attention instead (ablation) | base counting network · ShanghaiTech Part_A | MAE: 74.8 -> 84.5 | Hurt |
| add focus from global density (ablation) | base counting network · ShanghaiTech Part_A | MAE: 74.8 -> 71.7 | Improved |
| add both segmentation and global-density focus (ablation) | base counting network · ShanghaiTech Part_A | MAE: 74.8 -> 67.9 | Improved |
| add both focuses (ablation) | base counting network · WIDER FACE | MAE: 4.7 -> 3.2 | Improved |
| add CBAM combined attention instead of own focus (ablation) | base counting network · ShanghaiTech Part_A | MAE: 74.8 -> 73.5 | Mixed |
| replace geometry-adaptive kernel with proposed non-uniform kernel (ablation) | combined focus network · ShanghaiTech Part_A | MAE: 67.9 -> 65.2 | Improved |
| replace geometry-adaptive kernel with proposed kernel (ablation) | combined focus network · WIDER FACE | MAE: 4.2 -> 3.6 | Improved |
| add distiller module to encoder (ablation) | encoder-decoder counting network · ShanghaiTech Part_A | MAE: 114.8 -> 82.5 | Improved |
| add distiller between encoder and decoder (ablation) | encoder-decoder network · ShanghaiTech Part_A | MAE: 78.8 -> 74.8 | Improved |
| add the focus for free to Zhang et al. network (ablation) | Zhang et al. network · ShanghaiTech Part_A | MAE: 114.5 -> 110.1 | Improved |
| add the focus for free to Cao et al. network (ablation) | Cao et al. network · WIDER FACE | MAE: 8.5 -> 8.2 | Improved |
Closest papers by meaning
- Focus for Free in Density-Based Counting (IJCV 2024 · similarity 0.93)
- Learning to Count without Annotations (CVPR 2024 · similarity 0.89)
- DAVE - A Detect-and-Verify Paradigm for Low-Shot Counting (CVPR 2024 · similarity 0.87)
- A Novel Unified Architecture for Low-Shot Counting by Detection and Segmentation (NeurIPS 2024 · similarity 0.87)
- A Fixed-Point Approach to Unified Prompt-Based Counting (AAAI 2024 · similarity 0.86)
- Point Segment and Count: A Generalized Framework for Object Counting (CVPR 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.
