Research map / Image generation / Autoregressive and latent generation
EnvSocial-Diff: A Diffusion-Based Crowd Simulation Model with Environmental Conditioning and Individual-Group Interaction
Bingxue Zhao, Qi Zhang, Hui Huang
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 structured environmental conditioning and IGI module to SPDiff-style model | SPDiff · UCY | OT: 4.0564 -> 3.7292 | Improved |
| add structured environmental conditioning and IGI module | SPDiff · GC and UCY | collision count: GC 810 -> 906; UCY 372 -> 522 | Hurt |
| add obstacle features, then OOI features to the diffusion conditioning (ablation) | EnvSocial-Diff · UCY | MAE: 1.8597 -> 1.8337 -> 1.8271 | Improved |
| add lighting features (ablation) | EnvSocial-Diff · UCY | MMD and DTW: MAE 1.8271 -> 1.8182; MMD 0.0586 -> 0.0598 | Mixed |
| add explicit lighting module (ablation) | SPDiff · UCY | MAE: 1.8853 -> 1.8395 | Improved |
| add similarity terms sim1 and sim2 (without group conformity) to relative motion (ablation) | IGI module · GC | MAE: 0.8946 -> 0.9208 after adding sim2 without sim3 | Hurt |
| add all similarity terms including group conformity (ablation) | IGI module · UCY | MAE: 1.9055 -> 1.8182 | Improved |
| feed environmental conditioning into denoiser instead of post-hoc forces (ablation) | EnvSocial-Diff · UCY | MAE: 2.0764 -> 1.8182 | Improved |
| feed environmental conditioning into denoiser instead of post-hoc forces (ablation) | EnvSocial-Diff · GC | MAE: 0.8862 -> 0.8861 | No clear effect |
| add simple additive repulsive force (ablation) | EnvSocial-Diff · GC and UCY | collision count and MAE: UCY Col 522 -> 510; MAE 1.8182 -> 1.8609 | Mixed |
| use automatic GroundSAM detections instead of curated annotations (ablation) | EnvSocial-Diff · UCY | MAE: 1.8182 -> 1.9083 | Hurt |
| add environmental cues and IGI module to SPDiff on the full GC scene | SPDiff · Full GC | OT: 4.6824 -> 4.1334 | Improved |
Closest papers by meaning
- Social Physics Informed Diffusion Model for Crowd Simulation (AAAI 2024 · similarity 0.89)
- Continuous Locomotive Crowd Behavior Generation (CVPR 2025 · similarity 0.86)
- Learning Pedestrian Group Representations for Multi-modal Trajectory Prediction (ECCV 2022 · similarity 0.86)
- SocialCVAE: Predicting Pedestrian Trajectory via Interaction Conditioned Latents (AAAI 2024 · similarity 0.85)
- SocialCircle: Learning the Angle-based Social Interaction Representation for Pedestrian Trajectory Prediction (CVPR 2024 · similarity 0.84)
- Higher-order Relational Reasoning for Pedestrian Trajectory Prediction (CVPR 2024 · 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.
