Research map / Image generation / Image restoration
Counting Crowds in Bad Weather
Zhi-Kai Huang, Wei-Ting Chen, Yuan-Chun Chiang, Sy-Yen Kuo, Ming-Hsuan Yang
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 WACC weather-aware architecture (ablation) | baseline counter · JHU-Crowd++ (average) | MAE: 60.7 -> 60.1 | Improved |
| add input-dependent weather queries (ablation) | WACC · JHU-Crowd++ (average) | MAE: 60.1 -> 59.3 | Improved |
| add contrastive loss L_Con (ablation) | WACC + IWQ · JHU-Crowd++ (average) | MAE: 59.3 -> 53.1 | Improved |
| add L_CP loss (ablation) | WACC + IWQ + L_Con · JHU-Crowd++ (average) | MAE: 53.1 -> 52.3 | Improved |
| learn weather queries without explicit weather labels (ablation) | AWCC-Net · JHU-Crowd++ (average) | MAE: 56.1 -> 52.3 | Improved |
| add two-stage restoration (Unified) before counting | MAN · JHU-Crowd++ (adverse weather) | MAE: 105.3 -> 100.7 | Improved |
| add two-stage restoration (Unified) before counting | MAN · JHU-Crowd++ (clear) | MAE: 46.5 -> 56.9 | Hurt |
| add two-stage restoration (Unified) before counting | BL · JHU-Crowd++ (adverse weather) | MAE: 140.1 -> 134.0 | Improved |
| add restoration and finetune counter on restored images | GL · JHU-Crowd++ (adverse weather) | MAE: 115.9 -> 127.6 | Hurt |
| add restoration and finetune counter on restored images | MAN · JHU-Crowd++ (adverse weather) | MAE: 105.3 -> 117.1 | Hurt |
| use AWCC-Net instead of MAN for cross-dataset transfer | MAN · ShanghaiTech A (trained on JHU-Crowd++) | MAE: 81.2 -> 73.2 | Improved |
| use AWCC-Net instead of MAN for cross-dataset transfer | MAN · UCF-QNRF (trained on JHU-Crowd++) | MAE: 132.8 -> 120.9 | Improved |
Closest papers by meaning
- Counting without Scale: Scale-Consistent Error Correction for Crowd Counting (NeurIPS 2026 · similarity 0.84)
- Pushing the Frontiers of Unconstrained Crowd Counting: New Dataset and Benchmark Method (ICCV 2019 · similarity 0.83)
- JHU-CROWD++: Large-Scale Crowd Counting Dataset and A Benchmark Method (TPAMI 2020 · similarity 0.83)
- DISCount: Counting in Large Image Collections with Detector-Based Importance Sampling (AAAI 2024 · similarity 0.83)
- Composition Loss for Counting, Density Map Estimation and Localization in Dense Crowds (ECCV 2018 · similarity 0.83)
- See What We Cannot See: A Geo-guided Reasoning Benchmark for Object Counting under Adverse Earth Observation Conditions (CVPR 2026 · similarity 0.82)
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.
