agrimgupta92/sgan
Code for "Social GAN: Socially Acceptable Trajectories with Generative Adversarial Networks", Gupta et al, CVPR 2018
Combines an encoder-decoder RNN with a novel spatial pooling mechanism to model inter-agent interactions, enabling multimodal trajectory prediction that respects social conventions in crowded scenes. The architecture uses a generator-discriminator framework where the discriminator classifies real vs. predicted sequences to enforce socially plausible motion patterns. Implemented in PyTorch with pretrained models available for five benchmark datasets (ETH, UCY, etc.) and supports end-to-end training on custom trajectory data.
905 stars. No commits in the last 6 months.
Stars
905
Forks
272
Language
Python
License
MIT
Category
Last pushed
Nov 24, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/agrimgupta92/sgan"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Related frameworks
StanfordASL/Trajectron
Code accompanying "The Trajectron: Probabilistic Multi-Agent Trajectory Modeling with Dynamic...
StanfordASL/Trajectron-plus-plus
Code accompanying the ECCV 2020 paper "Trajectron++: Dynamically-Feasible Trajectory Forecasting...
uber-research/LaneGCN
[ECCV2020 Oral] Learning Lane Graph Representations for Motion Forecasting
jiachenli94/Awesome-Interaction-Aware-Trajectory-Prediction
A selection of state-of-the-art research materials on trajectory prediction
devendrachaplot/Neural-SLAM
Pytorch code for ICLR-20 Paper "Learning to Explore using Active Neural SLAM"