Saved in:
Bibliographic Details
Main Authors: Yang, Ruining, Xu, Yi, Fu, Yun, Su, Lili
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2409.17385
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908897831288832
author Yang, Ruining
Xu, Yi
Fu, Yun
Su, Lili
author_facet Yang, Ruining
Xu, Yi
Fu, Yun
Su, Lili
contents Trajectory prediction in autonomous driving has traditionally been studied from a model-centric perspective. However, existing datasets exhibit a strong long-tail distribution in scenario density, where common low-density cases dominate and safety-critical high-density cases are severely underrepresented. This imbalance limits model robustness and hides failure modes when standard evaluations average errors across all scenarios. We revisit trajectory prediction from a data-centric perspective and present Den-TP, a framework for density-aware dataset curation and evaluation. Den-TP first partitions data into density-conditioned regions using agent count as a dataset-agnostic proxy for interaction complexity. It then applies a gradient-based submodular selection objective to choose representative samples within each region while explicitly rebalancing across densities. The resulting subset reduces the dataset size by 50\% yet preserves overall performance and significantly improves robustness in high-density scenarios. We further introduce density-conditioned evaluation protocols that reveal long-tail failure modes overlooked by conventional metrics. Experiments on Argoverse 1 and 2 with state-of-the-art models show that robust trajectory prediction depends not only on data scale, but also on balancing scenario density.
format Preprint
id arxiv_https___arxiv_org_abs_2409_17385
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Den-TP: A Density-Balanced Data Curation and Evaluation Framework for Trajectory Prediction
Yang, Ruining
Xu, Yi
Fu, Yun
Su, Lili
Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
Trajectory prediction in autonomous driving has traditionally been studied from a model-centric perspective. However, existing datasets exhibit a strong long-tail distribution in scenario density, where common low-density cases dominate and safety-critical high-density cases are severely underrepresented. This imbalance limits model robustness and hides failure modes when standard evaluations average errors across all scenarios. We revisit trajectory prediction from a data-centric perspective and present Den-TP, a framework for density-aware dataset curation and evaluation. Den-TP first partitions data into density-conditioned regions using agent count as a dataset-agnostic proxy for interaction complexity. It then applies a gradient-based submodular selection objective to choose representative samples within each region while explicitly rebalancing across densities. The resulting subset reduces the dataset size by 50\% yet preserves overall performance and significantly improves robustness in high-density scenarios. We further introduce density-conditioned evaluation protocols that reveal long-tail failure modes overlooked by conventional metrics. Experiments on Argoverse 1 and 2 with state-of-the-art models show that robust trajectory prediction depends not only on data scale, but also on balancing scenario density.
title Den-TP: A Density-Balanced Data Curation and Evaluation Framework for Trajectory Prediction
topic Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2409.17385