Research map / Label-efficient and robust learning / Semi-supervised learning
Active Crowd Counting with Limited Supervision
Zhen Zhao, Miaojing Shi, Xiaoxiao Zhao, Li Li
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 PSSW partition-based selection instead of random selection (ablation) | active crowd counting · ShanghaiTech A (M=40, m=10) | MAE: 93.8 -> 85.4 | Improved |
| use PSSW instead of random selection (ablation) | active crowd counting · ShanghaiTech B (M=40, m=10) | MAE: 17.9 -> 14.6 | Improved |
| use PSSW instead of random selection (ablation) | active crowd counting · UCF_CC_50 (M=10, m=3) | MAE: 444.7 -> 387.3 | Improved |
| use Jenks-based partition instead of even partition (ablation) | PSSW · ShanghaiTech A (M=40, m=10) | MAE: 89.6 -> 84.4 | Improved |
| use GDSIM grid dissimilarity instead of global count difference (ablation) | PSSW · ShanghaiTech A (M=40, m=10) | MAE: 86.6 -> 84.4 | Improved |
| use GDSIM instead of global count difference (ablation) | PSSW · ShanghaiTech B (M=40, m=10) | MAE: 15.3 -> 14.4 | Improved |
| add gradient reversal layer distribution classifier (ablation) | PSSW · ShanghaiTech A (M=40, m=10) | MAE: 85.4 -> 82.7 | Improved |
| add latent MixUp on top of GRL (ablation) | PSSW + GRL · ShanghaiTech A (M=40, m=10) | MAE: 82.7 -> 80.4 | Improved |
| add GRL and latent MixUp (ablation) | PSSW · ShanghaiTech B (M=40, m=10) | MAE: 14.6 -> 12.7 | Improved |
| add GRL and latent MixUp (ablation) | PSSW · UCF_CC_50 (M=10, m=3) | MAE: 387.3 -> 351.4 | Improved |
| add GRL and latent MixUp to random selection (ablation) | RS · ShanghaiTech A (M=40, m=10) | MAE: 93.8 -> 87.3 | Improved |
| reduce m from 10 to 5 (ablation) | PSSW · ShanghaiTech A (M=30) | MAE: 93.5 -> 92.6 | Improved |
Closest papers by meaning
- Completely Self-Supervised Crowd Counting via Distribution Matching (ECCV 2022 · similarity 0.92)
- Reducing Spatial Labeling Redundancy for Active Semi-Supervised Crowd Counting (TPAMI 2022 · similarity 0.91)
- Leveraging Unlabeled Data for Crowd Counting by Learning to Rank (CVPR 2018 · similarity 0.91)
- Crowd Counting With Partial Annotations in an Image (ICCV 2021 · similarity 0.90)
- Weakly-Supervised Crowd Counting Learns from Sorting rather than Locations (ECCV 2020 · similarity 0.90)
- Almost Unsupervised Learning for Dense Crowd Counting (AAAI 2019 · similarity 0.90)
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.
