Research map / Detection and segmentation / Open-set and few-shot detection
From Open Set to Closed Set: Supervised Spatial Divide-and-Conquer for Object Counting
Haipeng Xiong, Hao Lü, Chengxin Liu, Liang Liu, Chunhua Shen, Zhiguo Cao
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 spatial divide-and-conquer to classification baseline (ablation) | Classification counter · Cross-dataset B->QNRF | MAE: 281.7 -> 166.5 | Improved |
| add S-DC to classification baseline (ablation) | Classification counter · Cross-dataset QNRF->A | MAE: 134.7 -> 61.8 | Improved |
| add S-DC to classification baseline (ablation) | Classification counter · Cross-dataset B->A | MAE: 179.4 -> 130.0 | Improved |
| add S-DC to regression baseline (ablation) | Regression counter · Cross-dataset B->QNRF | MAE: 230.3 -> 222.1 | Improved |
| add S-DC to regression baseline (ablation) | Regression counter · Cross-dataset A->B | MAE: 23.6 -> 22.9 | Improved |
| add S-DC to regression baseline (ablation) | Regression counter · Cross-dataset B->A | MAE: 133.9 -> 137.1 | Hurt |
| replace spatial attention with S-DC division (ablation) | SS-DCNet · ShanghaiTech Part_A | MAE: 64.1 MAE and 109.9 MSE | Improved |
| add learned upsampling loss L_up (ablation) | SS-DCNet (reg/cls) · ShanghaiTech Part_A | MAE: MAE reduces by 3.2/0.5 | Improved |
| add division consistency loss L_eq (ablation) | SS-DCNet (reg) · ShanghaiTech Part_A | MAE: 1.1 improvement in MAE | Improved |
| increase S-DC division stages beyond two (ablation) | SS-DCNet · ShanghaiTech Part_A | MAE | No clear effect |
| add explicit S-DC supervision (SS-DCNet vs S-DCNet) with resizing | SS-DCNet (cls) · UCF-QNRF | MAE: 15.8 in MAE | Improved |
| add explicit S-DC supervision (SS-DCNet vs S-DCNet) | SS-DCNet (reg) · UCF-QNRF | MAE: 5.3 in MAE | Improved |
Closest papers by meaning
- From Open Set to Closed Set: Counting Objects by Spatial Divide-and-Conquer (ICCV 2019 · similarity 0.97)
- Rethinking Spatial Invariance of Convolutional Networks for Object Counting (CVPR 2022 · similarity 0.89)
- Counting With Focus for Free (ICCV 2019 · similarity 0.88)
- Focus for Free in Density-Based Counting (IJCV 2024 · similarity 0.88)
- Learning to Count without Annotations (CVPR 2024 · similarity 0.87)
- CountGD: Multi-Modal Open-World Counting (NeurIPS 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.
