PPT: Pretraining with Pseudo-Labeled Trajectories for Motion Forecasting
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912928254394368 |
|---|---|
| author | Xu, Yihong Yin, Yuan Zablocki, Éloi Vu, Tuan-Hung Boulch, Alexandre Cord, Matthieu |
| author_facet | Xu, Yihong Yin, Yuan Zablocki, Éloi Vu, Tuan-Hung Boulch, Alexandre Cord, Matthieu |
| contents | Accurately predicting how agents move in dynamic scenes is essential for safe autonomous driving. State-of-the-art motion forecasting models rely on datasets with manually annotated or post-processed trajectories. However, building these datasets is costly, generally manual, hard to scale, and lacks reproducibility. They also introduce domain gaps that limit generalization across environments. We introduce PPT (Pretraining with Pseudo-labeled Trajectories), a simple and scalable pretraining framework that uses unprocessed and diverse trajectories automatically generated from off-the-shelf 3D detectors and tracking. Unlike data annotation pipelines aiming for clean, single-label annotations, PPT is a pretraining framework embracing off-the-shelf trajectories as useful signals for learning robust representations. With optional finetuning on a small amount of labeled data, models pretrained with PPT achieve strong performance across standard benchmarks, particularly in low-data regimes, and in cross-domain, end-to-end, and multi-class settings. PPT is easy to implement and improves generalization in motion forecasting. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_06491 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | PPT: Pretraining with Pseudo-Labeled Trajectories for Motion Forecasting Xu, Yihong Yin, Yuan Zablocki, Éloi Vu, Tuan-Hung Boulch, Alexandre Cord, Matthieu Computer Vision and Pattern Recognition Robotics Accurately predicting how agents move in dynamic scenes is essential for safe autonomous driving. State-of-the-art motion forecasting models rely on datasets with manually annotated or post-processed trajectories. However, building these datasets is costly, generally manual, hard to scale, and lacks reproducibility. They also introduce domain gaps that limit generalization across environments. We introduce PPT (Pretraining with Pseudo-labeled Trajectories), a simple and scalable pretraining framework that uses unprocessed and diverse trajectories automatically generated from off-the-shelf 3D detectors and tracking. Unlike data annotation pipelines aiming for clean, single-label annotations, PPT is a pretraining framework embracing off-the-shelf trajectories as useful signals for learning robust representations. With optional finetuning on a small amount of labeled data, models pretrained with PPT achieve strong performance across standard benchmarks, particularly in low-data regimes, and in cross-domain, end-to-end, and multi-class settings. PPT is easy to implement and improves generalization in motion forecasting. |
| title | PPT: Pretraining with Pseudo-Labeled Trajectories for Motion Forecasting |
| topic | Computer Vision and Pattern Recognition Robotics |
| url | https://arxiv.org/abs/2412.06491 |