Research map / Label-efficient and robust learning / Semi-supervised learning
Exclusivity-Guided Mask Learning for Semi-Supervised Crowd Instance Segmentation and Counting
Jiyang Huang, Hongru Chen, Wei Lin, Jia Wan, Antoni Chan
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 maskhead discriminative loss instead of Dice loss (ablation) | XMask · ShTech A (5%) | IoU: 0.2381 -> 0.3119 | Improved |
| add maskhead discriminative loss instead of Focal loss (ablation) | XMask · ShTech A (5%) | IoU: 0.2893 -> 0.3119 | Improved |
| add depthwise Gaussian smoothing (K=7, sigma=3) (ablation) | XMask · ShTech A | IoU: 0.3050 -> 0.3130 | Improved |
| add differentiable center sampling (ablation) | XMask · UCF-QNRF (5%) | IoU: 0.2806 -> 0.2807 | No clear effect |
| add differentiable center sampling (ablation) | XMask · UCF-QNRF (40%) | IoU: 0.2955 -> 0.2956 | No clear effect |
| add NNEC constraint to XMask (ablation) | XMask · UCF-QNRF (40%) | IoU@0.5: 0.278 -> 0.297 | Improved |
| add NNEC constraint to XMask (ablation) | XMask · ShTech A (5%) | IoU@0.5: 0.314 -> 0.313 | No clear effect |
| add NNEC constraint to FastSAM (ablation) | FastSAM · ShTech A (5%) | IoU@0.5: 0.279 -> 0.287 | Improved |
| unfreeze decoder only rather than head and backbone (ablation) | counting stage · ShTech A (5%) | MAE: 68.1 -> 64.6 | Improved |
| unfreeze decoder only rather than head only (ablation) | counting stage · ShTech A (5%) | MAE: 70.2 -> 64.6 | Improved |
| use XMask mask-prior framework over P2R pseudo-labels | P2R · ShTech A (5%) | MAE: 69.9 -> 64.6 | Improved |
| use XMask mask-prior framework over P2R | P2R · UCF-QNRF (5%) | MSE: 182.5 -> 163.7 | Improved |
Closest papers by meaning
- Semi-Supervised Crowd Counting via Self-Training on Surrogate Tasks (ECCV 2020 · similarity 0.88)
- The Devil Is in the Points: Weakly Semi-Supervised Instance Segmentation via Point-Guided Mask Representation (CVPR 2023 · similarity 0.88)
- Reducing Spatial Labeling Redundancy for Active Semi-Supervised Crowd Counting (TPAMI 2022 · similarity 0.87)
- Crowd-SAM:SAM as a smart annotator for object detection in crowded scenes (ECCV 2024 · similarity 0.87)
- MaskDiff: Modeling Mask Distribution with Diffusion Probabilistic Model for Few-Shot Instance Segmentation (AAAI 2024 · similarity 0.85)
- Point-to-Region Loss for Semi-Supervised Point-Based Crowd Counting (CVPR 2025 · similarity 0.85)
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.
