notesum.ai
Published at October 22Geometric Graph Neural Network Modeling of Human Interactions in Crowded Environments
cs.CV
cs.AI
cs.ET
cs.LG
Released Date: October 22, 2024
Authors: Sara Honarvar1, Yancy Diaz-Mercado1
Aff.: 1Department of Mechanical Engineering, University of Maryland

| Method | ETH | HOTEL | UNIV | ZARA01 | ZARA02 | Average |
|---|---|---|---|---|---|---|
| social-LSTM (Alahi et al. (2016)) | 1.09/2.35 | 0.79/1.76 | 0.67/1.40 | 0.47/1.00 | 0.56/1.17 | 0.72/1.54 |
| social-GAN (Gupta et al. (2018)) | 0.81/1.52 | 0.72/1.61 | 0.60/1.26 | 0.34/0.69 | 0.42/0.84 | 0.58/1.18 |
| Social-BiGAT (Kosaraju et al. (2019)) | 0.69/1.29 | 0.49/1.01 | 0.55/1.32 | 0.30/0.62 | 0.36/0.75 | 0.48/1.00 |
| social-STGCNN (Mohamed et al. (2020)) | 0.64/1.11 | 0.49/0.85 | 0.44/0.79 | 0.34/0.53 | 0.30/0.48 | 0.44/0.75 |
| Ours (Kernel 1: Inverse of Norm) | ||||||
| - View (Graph 1- Kernel 1) | 0.64/1.09 | 0.41/0.63 | 0.45/0.85 | 0.35/0.59 | 0.30/0.5 | 0.43/0.73 |
| - View, thresh (Graph 2- Kernel 1) | 0.76/1.48 | 0.40/0.69 | 0.45/0.85 | 0.33/0.52 | 0.29/0.47 | 0.44/0.80 |
| - Approach (Graph 3- Kernel 1) | 0.82/1.29 | 0.39/0.58 | 0.46/0.868 | 0.34/0.53 | 0.30/0.50 | 0.46/0.75 |
| - View, approach (Graph 4- Kernel 1) | 0.75/1.37 | 0.38/0.57 | 0.47/0.88 | 0.34/0.57 | 0.30/0.50 | 0.44/0.77 |
| Ours (Kernel 2: Exponential Decay) | ||||||
| - View (Graph 1- Kernel 2) | 0.80/1.54 | 0.47/0.77 | 0.46/0.84 | 0.37/0.58 | 0.43/0.67 | 0.51/0.88 |
| - View, thresh (Graph 2- Kernel 2) | 0.80/1.40 | 0.47/0.77 | 0.45/0.82 | 0.32/0.53 | 0.27/0.45 | 0.47/0.80 |
| - Approach (Graph 3- Kernel 2) | 0.86/1.37 | 0.34/0.47 | 0.47/0.86 | 0.36/0.56 | 0.28/0.46 | 0.47/0.74 |
| - View, approach (Graph 4- Kernel 2) | 0.76/1.36 | 0.47/0.83 | 0.48/0.88 | 0.33/0.52 | 0.30/0.48 | 0.47/0.81 |