PPT: Pre-Training with Pseudo-Labeled Trajectories for Motion Forecasting

• URL: http://arxiv.org/abs/2412.06491v1
• Date: Mon, 09 Dec 2024 13:48:15 GMT
• Title: PPT: Pre-Training with Pseudo-Labeled Trajectories for Motion Forecasting
• Authors: Yihong Xu, Yuan Yin, Tuan-Hung Vu, Alexandre Boulch, Éloi Zablocki, Matthieu Cord,
• Abstract summary: Motion forecasting for autonomous driving aims at anticipating trajectories of surrounding agents in complex urban scenarios.<n>In this work, we investigate a mixed strategy in MF training that first pre-train motion forecasters on pseudo-labeled data, then fine-tune them on annotated data.
• Score: 90.47748423913369
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Motion forecasting (MF) for autonomous driving aims at anticipating trajectories of surrounding agents in complex urban scenarios. In this work, we investigate a mixed strategy in MF training that first pre-train motion forecasters on pseudo-labeled data, then fine-tune them on annotated data. To obtain pseudo-labeled trajectories, we propose a simple pipeline that leverages off-the-shelf single-frame 3D object detectors and non-learning trackers. The whole pre-training strategy including pseudo-labeling is coined as PPT. Our extensive experiments demonstrate that: (1) combining PPT with supervised fine-tuning on annotated data achieves superior performance on diverse testbeds, especially under annotation-efficient regimes, (2) scaling up to multiple datasets improves the previous state-of-the-art and (3) PPT helps enhance cross-dataset generalization. Our findings showcase PPT as a promising pre-training solution for robust motion forecasting in diverse autonomous driving contexts.

Related papers

• Finetuning Pre-trained Model with Limited Data for LiDAR-based 3D Object Detection by Bridging Domain Gaps [8.897884780881535]
  LiDAR-based 3D object detectors often fail to adapt well to target domains with different sensor configurations. Recent studies suggest that pre-trained backbones can be learned in a self-supervised manner with large-scale unlabeled LiDAR frames. We propose a novel method, called Domain Adaptive Distill-Tuning (DADT), to adapt a pre-trained model with limited target data.
  arXiv  Detail & Related papers  (2024-10-02T08:22:42Z)
• TrajSSL: Trajectory-Enhanced Semi-Supervised 3D Object Detection [59.498894868956306]
  Pseudo-labeling approaches to semi-supervised learning adopt a teacher-student framework. We leverage pre-trained motion-forecasting models to generate object trajectories on pseudo-labeled data. Our approach improves pseudo-label quality in two distinct manners.
  arXiv  Detail & Related papers  (2024-09-17T05:35:00Z)
• Forecast-PEFT: Parameter-Efficient Fine-Tuning for Pre-trained Motion Forecasting Models [68.23649978697027]
  Forecast-PEFT is a fine-tuning strategy that freezes the majority of the model's parameters, focusing adjustments on newly introduced prompts and adapters. Our experiments show that Forecast-PEFT outperforms traditional full fine-tuning methods in motion prediction tasks. Forecast-FT further improves prediction performance, evidencing up to a 9.6% enhancement over conventional baseline methods.
  arXiv  Detail & Related papers  (2024-07-28T19:18:59Z)
• DPTraj-PM: Differentially Private Trajectory Synthesis Using Prefix Tree and Markov Process [1.800676987432211]
  We propose DPTraj-PM, a method to synthesize trajectory dataset under the differential privacy framework. DPTraj-PM discretizes the raw trajectories into neighboring cells, and models them by combining a prefix tree structure and an m-order Markov process. The output traces crafted by DPTraj-PM not only preserves the patterns and variability in individuals' mobility behaviors, but also protects individual privacy.
  arXiv  Detail & Related papers  (2024-04-22T11:52:11Z)
• Unsupervised Domain Adaptation for Self-Driving from Past Traversal Features [69.47588461101925]
  We propose a method to adapt 3D object detectors to new driving environments. Our approach enhances LiDAR-based detection models using spatial quantized historical features. Experiments on real-world datasets demonstrate significant improvements.
  arXiv  Detail & Related papers  (2023-09-21T15:00:31Z)
• SPOT: Scalable 3D Pre-training via Occupancy Prediction for Learning Transferable 3D Representations [76.45009891152178]
  Pretraining-finetuning approach can alleviate the labeling burden by fine-tuning a pre-trained backbone across various downstream datasets as well as tasks. We show, for the first time, that general representations learning can be achieved through the task of occupancy prediction. Our findings will facilitate the understanding of LiDAR points and pave the way for future advancements in LiDAR pre-training.
  arXiv  Detail & Related papers  (2023-09-19T11:13:01Z)
• Pre-training on Synthetic Driving Data for Trajectory Prediction [61.520225216107306]
  We propose a pipeline-level solution to mitigate the issue of data scarcity in trajectory forecasting. We adopt HD map augmentation and trajectory synthesis for generating driving data, and then we learn representations by pre-training on them. We conduct extensive experiments to demonstrate the effectiveness of our data expansion and pre-training strategies.
  arXiv  Detail & Related papers  (2023-09-18T19:49:22Z)
• Efficient Baselines for Motion Prediction in Autonomous Driving [7.608073471097835]
  Motion Prediction (MP) of multiple surroundings agents is a crucial task in arbitrarily complex environments. We aim to develop compact models using State-Of-The-Art (SOTA) techniques for MP, including attention mechanisms and GNNs.
  arXiv  Detail & Related papers  (2023-09-06T22:18:16Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.