List your company
ResearchFormerly WizWand

Trajectory Prediction on ZARA1

0.16Min ADE (H=20)FEP-Diff
0.15240.20370.2550.3063Mar 20, 2023Sep 29, 2023Apr 10, 2024Oct 21, 2024May 2, 2025Nov 12, 2025May 25, 2026
Updated 5mo ago

Evaluation results

MethodLinks
FEP-DiffVenue=-, K=20, Observability=agent-centric2026.050.16——0.3——
NPSN2023.030.17——0.31——
MID2023.030.17——0.3——
MID2024.070.17——0.3——
SocialVAE2024.070.17——0.29——
PPT2024.070.17——0.3——
GP-GraphVenue=ECCV’22, K=20, Observability=agent-centric2026.050.17——0.31——
MoFlowVenue=CVPR’25, K=20, Observability=agent-centric2026.050.17——0.3——
Agentformer2023.030.18——0.3——
MemoNet2023.030.18——0.32——
LED2023.030.18——0.26——
AgentFormer2024.070.18——0.3——
MemoNet2024.070.18——0.32——
LED2024.070.18——0.26——
NPSN2024.070.18——0.32——
TUTR2024.070.18——0.34——
AgentFormerVenue=ICCV’21, K=20, Observability=agent-centric2026.050.18——0.3——
NPSNVenue=CVPR’22, K=20, Observability=agent-centric2026.050.18——0.32——
SocialVAEVenue=ECCV’22, K=20, Observability=agent-centric2026.050.18——0.3——
EqMotionVenue=CVPR’23, K=20, Observability=agent-centric2026.050.18——0.32——
EigenTrajectory2024.070.19——0.33——
SingularTrajVenue=CVPR’24, K=20, Observability=agent-centric2026.050.19——0.32——
EigenTrajectoryVenue=ICCV’23, K=20, Observability=agent-centric2026.050.2——0.35——
GroupNet2023.030.21——0.39——
PCCSNet2024.070.21——0.44——
GroupnetVenue=CVPR’22, K=20, Observability=agent-centric2026.050.21——0.39——
PECNet2023.030.22——0.39——
PECNet2024.070.22——0.39——
MIDVenue=CVPR’22, K=20, Observability=agent-centric2026.050.22——0.45——
Trajectron++2023.030.24——0.42——
STAR2023.030.29——0.52——
STAR2024.070.29——0.52——
Trajectron++Venue=ECCV’20, K=20, Observability=agent-centric2026.050.29——0.43——
NMMP2023.030.32——0.66——
Social-GAN2023.030.35——0.68——
Social-GAN2024.070.35——0.68——
AutoTrajectoryzero-shot=true2024.03—0.480.91———
Causal-STGAT2022.03—0.320.64———
DD-MDNEvaluation Protocol=Momentary Observation, Input Horizon=2, K=202026.02————0.170.28
EigenTrajEvaluation Protocol=Momentary Observation, Input Horizon=2, K=202026.02————0.190.35
Kalman filterzero-shot=true2024.03—0.450.98———
Linearzero-shot=true2024.03—0.430.96———
LMTraj-ZEROFoundation Model=GPT-3.5, zero-shot=true2024.03—0.470.91———
LMTraj-ZEROFoundation Model=GPT-4, zero-shot=true2024.03—0.330.66———
MATFsampling=Best-of-202022.06—0.260.45———
MIDEvaluation Protocol=Momentary Observation, Input Horizon=2, K=202026.02————0.30.56
NCE-Trajectron++2022.03—0.220.41———
NPSN-Causal-STGATsampling_strategy=NPSN2022.03—0.320.65———
NPSN-NCE-Trajectron++sampling_strategy=NPSN2022.03—0.190.35———
NPSN-PECNetsampling_strategy=NPSN2022.03—0.210.33———
NPSN-SGANsampling_strategy=NPSN2022.03—0.340.68———
NPSN-SGCNsampling_strategy=NPSN2022.03—0.180.32———
NPSN-STGATsampling_strategy=NPSN2022.03—0.340.68———
NPSN-STGCNNsampling_strategy=NPSN2022.03—0.250.43———
NPSN-Trajectron++sampling_strategy=NPSN2022.03—0.190.36———
PECNet2022.03—0.250.45———
S-ATTNmode=most likely2022.06—0.20.52———
S-GANsampling=Best-of-202022.06—0.340.69———
S-LSTMmode=most likely2022.06—0.471———
ScePTsampling=Best-of-202022.06—0.130.77———
ScePTmode=most likely2022.06—0.181.1———
SGCN2022.03—0.290.51———
SingTrajEvaluation Protocol=Momentary Observation, Input Horizon=2, K=202026.02————0.190.33
Social-GAN2022.03—0.350.68———
Social-STGCNN2022.03—0.340.53———
SoPhiesampling=Best-of-202022.06—0.30.63———
STGAT2022.03—0.340.69———
Stopzero-shot=true2024.03—2.514.61———
Trajectron++2022.03—0.240.41———
Trajectron++sampling=Best-of-202022.06—0.150.33———
Trajectron++mode=most likely2022.06—0.30.79———