The invention relates to the field of trajectory prediction, in particular to a long-tail pedestrian trajectory prediction method based on subclass balance contrast learning. The method comprises the steps that a training data set is constructed, the training data set comprises multiple sets of track data, and each set of track data comprises a historical track X and a future real track Y; constructing a prediction model, wherein the prediction model comprises a motion feature coding module, a subclass balance contrast learning module, a transformer encoder and a classifier; the subclass balance comparison learning module divides the head class into a plurality of subclasses of which the sample size is similar to the average value of the sample size of the tail class, and carries out comparison learning according to subclass labels; training a prediction model by using the training data set to obtain a trained prediction model; and inputting the current historical trajectory into the trained prediction model, performing prediction by using the trained prediction model, and outputting a future prediction trajectory. According to the method, category balance and instance balance can be realized, the long-tail distribution problem of pedestrian trajectory data is effectively solved, and generalization in different scenes is improved.
本发明涉及轨迹预测领域,具体涉及一种基于子类平衡对比学习的长尾行人轨迹预测方法。方法包括:构建训练数据集,训练数据集包括多组轨迹数据,每组轨迹数据包括历史轨迹X和未来真实轨迹Y;构建预测模型,预测模型包括运动特征编码模块、子类平衡对比学习模块、transformer编码器和分类器;子类平衡对比学习模块将头部类分为样本量与尾部类样本量均值相近的多个子类,依据子类标签进行对比学习;使用训练数据集训练预测模型,得到训练好的预测模型;将当前历史轨迹输入训练好的预测模型,利用训练好的预测模型进行预测,输出未来预测轨迹。本发明可以实现类别平衡和实例平衡,有效应对行人轨迹数据的长尾分布问题并提高在不同场景下的泛化性。
Long-tail pedestrian trajectory prediction method based on subclass balance contrast learning
基于子类平衡对比学习的长尾行人轨迹预测方法
2024-11-29
Patent
Electronic Resource
Chinese
IPC: | G06T Bilddatenverarbeitung oder Bilddatenerzeugung allgemein , IMAGE DATA PROCESSING OR GENERATION, IN GENERAL / B60W CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION , Gemeinsame Steuerung oder Regelung von Fahrzeug-Unteraggregaten verschiedenen Typs oder verschiedener Funktion / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / G06V |
PEDESTRIAN TRAJECTORY PREDICTION WITH LEARNING-BASED APPROACHES: A COMPARATIVE STUDY
British Library Conference Proceedings | 2019
|Crossing pedestrian trajectory prediction method and device
European Patent Office | 2022
|