Optical motion capture data processing method, model training method and device

The invention provides an optical motion capture data processing method and device. Relates to the technical field of computers. The method includes obtaining first mark point position frame sequences, wherein each mark point position frame sequence comprises a plurality of mark points, and one mark...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: WANG YUPAN, CHEN KANG, ZHANG WEIDONG
Format: Patent
Sprache:chi ; eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The invention provides an optical motion capture data processing method and device. Relates to the technical field of computers. The method includes obtaining first mark point position frame sequences, wherein each mark point position frame sequence comprises a plurality of mark points, and one mark point is used for marking one or more joint positions; based on a target prediction model and the first mark point position frame sequence, performing prediction to obtain low-dimension prediction data; and based on a target decoding model and the prediction data, performing decoding to obtain target skeletal animation data. Clean skeleton animation data corresponding to the mark point position frame sequence can be directly predicted from the mark point position frame sequence containing errors and noise, and the data processing efficiency is improved. 本发明提供了一种光学运动捕捉数据处理方法及装置。涉及计算机技术领域。通过获取第一标记点位置帧序列,每个标记点位置帧序列包括多个标记点,一个所述标记点用于标记一个或多个关节位置;基于目标预测模型以及所述第一标记点位置帧序列,预测得到低维度的预测数据;基于目标解码模型以及所述预测数据,解码得到目标骨骼动画数据。可以直接从包含错误