Trajectory Prediction on Stanford Drone (5-fold cross val)
0.7Error @ 1secShotgun
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ShotgunSampling Strategy=Top 10%, Visual Input=None, Resolution=1/52019.08 | 0.7 | 1.7 | 3 | 4.5 | 91.6 | |
| CF-VAE + pRSampling Strategy=Top 10%, C=0.2, Visual Input=None, Resolution=1/52019.08 | 0.7 | 1.5 | 2.5 | 3.6 | 84.6 | |
| CF-VAE + pRSampling Strategy=Top 10%, C=0.2, Visual Input=RGB, Resolution=1/52019.08 | 0.7 | 1.5 | 2.4 | 3.5 | 84.1 | |
| BMS-CVAESampling Strategy=Top 10%, Visual Input=RGB, Resolution=1/52019.08 | 0.8 | 1.7 | 3.1 | 4.6 | 126.6 | |
| MoG-CVAESampling Strategy=Top 10%, M=3, Visual Input=None, Resolution=1/52019.08 | 0.8 | 1.7 | 2.7 | 3.9 | 86.1 | |
| CF-VAESampling Strategy=Top 10%, Regularization=None, Visual Input=None, Resolution=1/52019.08 | 0.9 | 1.9 | 3.3 | 4.7 | 96.2 | |
| DESIRE-SI-IT4Sampling Strategy=Top 10%, Visual Input=RGB, Resolution=1/52019.08 | 1.2 | 2.3 | 3.4 | 5.3 | — | |
| STCNNSampling Strategy=Top 10%, Visual Input=RGB, Resolution=1/52019.08 | 1.2 | 2.1 | 3.3 | 4.6 | — |