Research map / 3D vision / Depth and stereo
Stacked Homography Transformations for Multi-View Pedestrian Detection
Liangchen Song, Jialian Wu, Ming Yang, Qian Zhang, Yuan Li, Junsong Yuan
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 |
|---|---|---|---|
| replace MVDet ground-plane feature projection with SHOT | MVDet · WILDTRACK | MODA: 88.2 -> 90.2 | Improved |
| replace MVDet ground-plane feature projection with SHOT | MVDet · MultiviewX | MODA: 83.9 -> 88.3 | Improved |
| apply SHOT to image-level projection | MVDet · MultiviewX (project images) | MODA: 19.5 -> 32.3 | Improved |
| apply SHOT to result-level projection | MVDet · MultiviewX (project results) | MODA: 73.2 -> 77.0 | Improved |
| remove the large kernel in the final occupancy classification layer (ablation) | MVDet · MultiviewX | MODA: 83.9 -> 77.2 | Hurt |
| remove the large kernel in the final occupancy classification layer (ablation) | SHOT · MultiviewX | MODA: 88.3 -> 86.1 | Hurt |
| increase stack depth D from 1 to 4 (ablation) | SHOT · MultiviewX | MODA: 85.5 -> 88.3 | Improved |
| increase stack depth D from 4 to 5 (ablation) | SHOT · MultiviewX | MODA: 88.3 -> 88.6 | No clear effect |
| increase stack depth D from 5 to 6 (ablation) | SHOT · MultiviewX | MODA: 88.6 -> OOM | Hurt |
| use SHOT when training and testing on different scenes and cameras | MVDet · MultiviewX (different scene and cameras for train and test) | MODA: 33.0 -> 49.1 | Improved |
| use SHOT when training and testing on different scenes and cameras | MVDet · MultiviewX (different scene and cameras for train and test) | Precision: 64.5 -> 73.3 | Improved |
Closest papers by meaning
- MV2GF: Multi-view Pedestrian Detection with a Visual Geometric Foundation Model (ECCV 2026 · similarity 0.87)
- Multi-View Pedestrian Occupancy Prediction with a Novel Synthetic Dataset (AAAI 2025 · similarity 0.85)
- Towards Generalizable Multi-Camera 3D Object Detection via Perspective Rendering (AAAI 2025 · similarity 0.85)
- Noisy One-point Homographies are Surprisingly Good (CVPR 2024 · similarity 0.85)
- Mask-Homo: Pseudo Plane Mask-Guided Unsupervised Multi-Homography Estimation (AAAI 2024 · similarity 0.84)
- WildCAT3D: Appearance-Aware Multi-View Diffusion in the Wild (NeurIPS 2025 · similarity 0.84)
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.
