Research map / Detection and segmentation / CNN features
JHU-CROWD++: Large-Scale Crowd Counting Dataset and A Benchmark Method
Vishwanath A. Sindagi, Rajeev Yasarla, Vishal M. Patel
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 residual learning (ablation) | VGG16 base network · JHU-CROWD++ val | MAE: 81.1 -> 77.5 (MSE 300.5 -> 290.6) | Improved |
| add confidence estimation blocks without uncertainty (lambda_c=0) (ablation) | Base + R (VGG16) · JHU-CROWD++ val | MAE: 77.5 -> 77.1 | No clear effect |
| add uncertainty guidance (lambda_c=1) to residual blocks (ablation) | Base + R (VGG16) · JHU-CROWD++ val | MAE: 77.1 -> 74.1 (MSE 290.5 -> 275.5) | Improved |
| add class-conditioning conv block without image-level loss (lambda_w=0) (ablation) | Base + R + UREB (VGG16) · JHU-CROWD++ val | MAE: 74.1 -> 74.6 | No clear effect |
| add image-level classification loss (lambda_w=0.01) to class-conditioning (ablation) | Base + R + UREB-C (VGG16) · JHU-CROWD++ val | MAE: 74.6 -> 67.9 (MSE 274.1 -> 262.1) | Improved |
| add residual learning (ablation) | Res101 base network · JHU-CROWD++ val | MAE: 72.1 -> 68.5 | Improved |
| add uncertainty guidance (lambda_c=1) to residual blocks (ablation) | Base + R (Res101) · JHU-CROWD++ val | MAE: 68.2 -> 62.5 | Improved |
| add class-conditioning with image-level loss (lambda_w=0.01) (ablation) | Base + R + UREB (Res101) · JHU-CROWD++ val | MAE: 62.5 -> 57.6 (MSE 258.1 -> 244.4) | Improved |
| use weather labels with multi-task learning only (ablation) | CG-DRCN · JHU-CROWD++ weather val | MAE: 78.4 -> 74.4 (MSE 170.5 -> 140.1) | Improved |
| add conditioning of residual estimation on weather labels over multi-task learning (ablation) | CG-DRCN · JHU-CROWD++ weather val | MAE: 74.4 -> 63.6 (MSE 140.1 -> 116.6) | Improved |
| add residual estimation branches at multiple scales (Y6 only to Y6+Y5+Y4+Y3) (ablation) | VGG16 · JHU-CROWD++ val | MAE: 81.1 -> 67.9 | Improved |
| add residual estimation branches at multiple scales (Y6 only to Y6+Y5+Y4+Y3) (ablation) | Res101 · JHU-CROWD++ val | MAE: 72.1 -> 57.6 | Improved |
Closest papers by meaning
- Pushing the Frontiers of Unconstrained Crowd Counting: New Dataset and Benchmark Method (ICCV 2019 · similarity 0.97)
- Residual Regression With Semantic Prior for Crowd Counting (CVPR 2019 · similarity 0.89)
- NWPU-Crowd: A Large-Scale Benchmark for Crowd Counting and Localization (TPAMI 2020 · similarity 0.89)
- Counting without Scale: Scale-Consistent Error Correction for Crowd Counting (NeurIPS 2026 · similarity 0.88)
- Composition Loss for Counting, Density Map Estimation and Localization in Dense Crowds (ECCV 2018 · similarity 0.88)
- Wide-Area Crowd Counting: Multi-view Fusion Networks for Counting in Large Scenes (IJCV 2022 · 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.
