Research map / Detection and segmentation / Open-set and few-shot detection
From Open Set to Closed Set: Counting Objects by Spatial Divide-and-Conquer
Haipeng Xiong, Hao Lu, Chengxin Liu, Liang Liu, Zhiguo Cao, Chunhua Shen
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 division decider (learned division mask) instead of softmax attention fusion (ablation) | S-DCNet · ShanghaiTech Part_A | MAE: 64.1 -> 58.3 | Improved |
| add division decider instead of softmax attention fusion (ablation) | S-DCNet · ShanghaiTech Part_A | MSE: 109.9 -> 95.0 | Improved |
| add spatial divide-and-conquer (S-DC) to classification (ablation) | VGG16 count classifier · ShanghaiTech Part_A | MAE | Improved |
| add S-DC to regression baseline (ablation) | VGG16 count regressor · ShanghaiTech Part_A | MAE | Improved |
| constrain regressor output to closed set [0,Cmax] with S-DC rather than open-set output (ablation) | regression + S-DC · ShanghaiTech Part_A | MAE | Improved |
| reformulate counting as classification (with S-DC) rather than regression (ablation) | S-DCNet · ShanghaiTech Part_A | MAE | Improved |
| classification without S-DC vs regression (ablation) | VGG16 count classifier · ShanghaiTech Part_A | MAE | Hurt |
| add merging loss L_R^2 on final division output (ablation) | S-DCNet · ShanghaiTech Part_A | MAE | Improved |
| lower Cmax to smaller quantiles (100% down to 80%) with S-DC (ablation) | S-DCNet · ShanghaiTech Part_A | MAE | No clear effect |
| lower Cmax quantile without S-DC (ablation) | classification baseline · ShanghaiTech Part_A | MAE | Hurt |
| use Two-Linear partition (finer step 0.05 in (0,0.5]) instead of One-Linear (ablation) | S-DCNet · ShanghaiTech Part_A | MAE | Improved |
| use two-stage S-DC instead of one stage (ablation) | S-DCNet · ShanghaiTech Part_A | MAE | Improved |
Closest papers by meaning
- From Open Set to Closed Set: Supervised Spatial Divide-and-Conquer for Object Counting (IJCV 2023 · similarity 0.97)
- Wide-Area Crowd Counting: Multi-view Fusion Networks for Counting in Large Scenes (IJCV 2022 · similarity 0.89)
- Rethinking Spatial Invariance of Convolutional Networks for Object Counting (CVPR 2022 · similarity 0.88)
- Learning Spatial Awareness to Improve Crowd Counting (ICCV 2019 · similarity 0.88)
- Learning To Count Everything (CVPR 2021 · similarity 0.88)
- Wide-Area Crowd Counting via Ground-Plane Density Maps and Multi-View Fusion CNNs (CVPR 2019 · similarity 0.88)
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.
