Egotraj Bench

EgoTraj-Bench is the first real-world benchmark for pedestrian trajectory prediction under ego-centric noisy observations. It pairs noisy first-person-view derived trajectories with clean bird's-eye-view ground truth, built upon the TBD dataset. The dataset was uploaded by ZoeyLIU1999 and last updated on April 22, 2026.

Use Cases

• Evaluating model robustness based on noisy first-person-view trajectory observations.
• Training trajectory prediction models based on paired noisy and clean trajectory data.
• Benchmarking algorithms for autonomous navigation based on pedestrian movement data.

Strengths

• First benchmark of its kind for real-world, ego-centric noisy observations.
• Provides paired data: noisy first-person-view trajectories with clean bird's-eye-view ground truth.
• Designed for robust evaluation under deployment-realistic conditions.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file formats are unknown, which may limit suitability assessment.

Provenance

Source

Built upon the TBD dataset.

Collection Method

Pairs noisy first-person-view derived trajectories with clean bird's-eye-view ground truth.

Freshness

Last updated 2026-04-22 18:17:13; freshness should be verified.