MMASD+
收藏资源简介:
MMASD+数据集由特拉华大学创建,是一个用于自闭症谱系障碍儿童隐私保护行为分析的增强型多模态数据集。该数据集包含1315个视频片段,涵盖3D骨骼坐标、3D身体网格和光流数据等多种数据模态。数据集的创建过程涉及使用Yolov8和Deep SORT算法进行个体识别和跟踪,确保隐私保护的同时捕捉治疗过程中的运动特征。MMASD+数据集主要应用于自闭症研究领域,旨在通过分析儿童在治疗过程中的行为,监测治疗进展并制定个性化治疗策略。
The MMASD+ dataset, developed by the University of Delaware, is an enhanced multimodal dataset for privacy-preserving behavioral analysis of children with autism spectrum disorder (ASD). This dataset comprises 1,315 video clips, covering multiple data modalities including 3D skeletal coordinates, 3D body meshes, and optical flow data. The development of this dataset involved using the Yolov8 and Deep SORT algorithms for person identification and tracking, which ensures privacy protection while capturing motion features during therapeutic interventions. Primarily applied in the field of autism research, the MMASD+ dataset aims to analyze children’s behaviors during therapy to monitor treatment progress and develop personalized therapeutic strategies.




