Research map / Graph learning / Knowledge and temporal graphs
Social-STGCNN: A Social Spatio-Temporal Graph Convolutional Neural Network for Human Trajectory Prediction
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 recurrent trajectory modeling and feature aggregation with a spatio-temporal graph model and temporal convolutions | Social-STGCNN · ETH/UCY AVG | ADE: 0.45 -> 0.44 | Helped |
| Replace recurrent trajectory modeling and feature aggregation with a spatio-temporal graph model and temporal convolutions | Social-STGCNN · ETH/UCY AVG | FDE: 0.94 -> 0.75 | Helped |
| Use one ST-GCNN layer and five TXP-CNN layers (ablation) | Social-STGCNN · ETH/UCY | ADE / FDE: 0.44 / 0.75 | Helped |
| Increase the number of ST-GCNN layers from one to three, with five TXP-CNN layers (ablation) | Social-STGCNN · ETH/UCY | ADE / FDE: 0.44 / 0.75 -> 0.54 / 0.93 | Hurt |
| Increase the number of ST-GCNN layers from one to five, with five TXP-CNN layers (ablation) | Social-STGCNN · ETH/UCY | ADE / FDE: 0.44 / 0.75 -> 0.59 / 1.02 | Hurt |
| Increase the number of ST-GCNN layers from one to seven, with five TXP-CNN layers (ablation) | Social-STGCNN · ETH/UCY | ADE / FDE: 0.44 / 0.75 -> 0.62 / 1.07 | Hurt |
| Use seven rather than five TXP-CNN layers while keeping one ST-GCNN layer (ablation) | Social-STGCNN · ETH/UCY | ADE / FDE: 0.44 / 0.75 -> 0.48 / 0.87 | Hurt |
| Replace the proposed similarity kernel with an L2-distance kernel (ablation) | Social-STGCNN · ETH/UCY | ADE / FDE: 0.44 / 0.75 -> 0.48 / 0.84 | Mixed |
| Replace the proposed similarity kernel with an exponential distance kernel (ablation) | Social-STGCNN · ETH/UCY | ADE / FDE: 0.44 / 0.75 -> 0.50 / 0.84 | Mixed |
| Replace the proposed similarity kernel with inverse distance plus epsilon (ablation) | Social-STGCNN · ETH/UCY | ADE / FDE: 0.44 / 0.75 -> 0.48 / 0.88 | Mixed |
| Set all adjacency values between different nodes to one instead of using the proposed kernel (ablation) | Social-STGCNN · ETH/UCY | ADE / FDE: 0.44 / 0.75 -> 0.49 / 0.79 | Mixed |
Closest papers by meaning
- Recursive Social Behavior Graph for Trajectory Prediction (CVPR 2020 · similarity 0.92)
- Social-Transmotion: Promptable Human Trajectory Prediction (ICLR 2024 · similarity 0.91)
- A Unified Environmental Network for Pedestrian Trajectory Prediction (AAAI 2024 · similarity 0.91)
- SocialCVAE: Predicting Pedestrian Trajectory via Interaction Conditioned Latents (AAAI 2024 · similarity 0.90)
- Higher-order Relational Reasoning for Pedestrian Trajectory Prediction (CVPR 2024 · similarity 0.89)
- SocialCircle: Learning the Angle-based Social Interaction Representation for Pedestrian Trajectory Prediction (CVPR 2024 · 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.