Widar3.0-G6D

该数据集是对Widar3.0数据集的修改版本，专注于三种室内环境下16名参与者执行的6种手势动作。该数据集特别为跨领域的人类活动识别任务而定制。除了包含含有7个领域的训练子集和9个领域的测试子集外，数据预处理步骤还包括时间子采样和特征提取，以提高计算效率。原始数据集包含68,246个手势样本，预处理后样本数量为58,648个。该数据集的任务是进行人类活动识别。This dataset is a modified version of the Widar3.0 dataset, focusing on 6 gesture movements performed by 16 participants across three indoor environments. It is specifically customized for cross-domain human activity recognition tasks. In addition to including a training subset with 7 domains and a test subset with 9 domains, the data preprocessing steps encompass temporal subsampling and feature extraction to enhance computational efficiency. The original dataset contains 68,246 gesture samples, while the number of samples after preprocessing is 58,648. The task of this dataset is human activity recognition.

搜集汇总

背景与挑战

背景概述

Widar3.0-G6D数据集是基于Widar3.0数据集创建的特定子集，专门用于跨领域WiFi信号的人类活动识别研究。该数据集通过处理四个原始CSI文件生成，支持领域对抗测试时适应（DATTA）方法的训练和评估，旨在提升模型在不同环境下的泛化能力。