InhwanBae/Crowd-Behavior-Generation
Official Code for "Continuous Locomotive Crowd Behavior Generation (CVPR 2025)"
Combines a diffusion-based crowd emitter with a state-switching simulator to generate lifelong agent trajectories from single scene images, eliminating dependency on observation sequences. Trained on ETH, UCY, SDD, and EDIN datasets with PyTorch 2.2.2, it supports both synthetic and real-world scenarios via Sim2Real/Real2Sim evaluation, plus 3D visualization through CARLA integration for interactive scene population and behavior customization.
Stars
44
Forks
8
Language
Python
License
MIT
Category
Last pushed
Nov 07, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/InhwanBae/Crowd-Behavior-Generation"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Related models
InhwanBae/SingularTrajectory
Official Code for "SingularTrajectory: Universal Trajectory Predictor using Diffusion Model (CVPR 2024)"
XinGP/DGFNet
[RAL 2025]Multi-Agent Trajectory Prediction with Difficulty-Guided Feature Enhancement Network
DSL-Lab/MoFlow
[CVPR 2025] MoFlow: One-Step Flow Matching for Human Trajectory Forecasting via Implicit Maximum...
InhwanBae/GraphTERN
Official Code for "A Set of Control Points Conditioned Pedestrian Trajectory Prediction (AAAI 2023)"
tr3e/InterGen
[IJCV 2024] InterGen: Diffusion-based Multi-human Motion Generation under Complex Interactions