An EMG and IMU Dataset for the Italian Sign Language Alphabet

This data repository contains surface electromyography (EMG) and Inertial Measurement Unit (IMU) data collected with the Myo Gesture Control Armband about the gestures of the 26 letters of the Italian Sign Language Alphabet. The dataset contains 780 gesture samples (30 for each letter of the alphabet) and is organized into 26 directories, one for each letter of the alphabet. Each directory includes 30 json files, one for each sample of the gesture representing a letter. Each json file is named using a Global Unique Identifier (GUID), and include the following field: - timestamp, a string representing the date and time of the gesture acquisition. For example, the string “09/07/20/10:03:19” suggests that the gesture and its acquisition were performed the 9th of July 2020, at 10:03:19 a.m. - duration, an integer describing how long was the data acquisition of the gesture in milliseconds. The value is 2000 in all the json files, since the time window for the data acquisition was 2 seconds; - emg, an object representing the EMG data of the gesture. It has two fields - frequency, i.e. the sampling frequency (in Hz) of the values from the EMG sensors. This value is 200 in all the json files; - data, a 400 x 8 integer matrix. Each row is then an 8-dimensional array including the values from the 8 EMG sensors of the Myo Armband. Therefore, data is the time series of the values from the EMG sensors during the acquisition of the gesture; - imu, an object representing the IMU data of the gesture acquisition. It has two fields - frequency, i.e. the sampling frequency (in Hz) of the values from the IMU. This value is 200 in all the json files; - data, a 400 elements length object array. Each object has three fields, namely gyroscope (an array composed by 3 floating point values), acceleration (an array composed by 3 floating point values), and rotation (an array composed by 4 floating point values). The dataset can be used during the training and the testing of algorithms and techniques for automatic gesture recognition and, specifically, for automatic sign language interpretation.本数据集仓库收录了由Myo手势控制手环（Myo Gesture Control Armband）采集的表面肌电（surface electromyography, EMG）与惯性测量单元（Inertial Measurement Unit, IMU）数据，对应意大利手语字母表中26个字母的手势动作。该数据集共包含780个手势样本（每个字母对应30个样本），按26个目录组织，每个字母对应一个专属目录。每个目录下包含30个JSON文件，分别对应该字母手势的每个采集样本。每个JSON文件以全局唯一标识符（Global Unique Identifier, GUID）命名，包含以下字段： - timestamp（时间戳）：字符串格式，代表手势采集的日期与时间。示例字符串"09/07/20/10:03:19"表示该手势及采集操作于2020年7月9日上午10点03分19秒完成； - duration（采集时长）：整数类型，描述手势数据采集的总时长（单位：毫秒）。所有JSON文件中该值均为2000，因数据采集的时间窗口设定为2秒； - emg（肌电数据对象）：代表手势肌电采集数据的对象，包含两个子字段： - frequency（采样频率）：肌电传感器数据的采样频率（单位：赫兹，Hz），所有文件中该值均为200； - data（肌电数据矩阵）：为400×8的整数矩阵。每一行均为8维数组，对应Myo手环8个肌电传感器的实时采集值，因此该矩阵为手势采集期间肌电传感器数据的完整时间序列； - imu（惯性测量单元数据对象）：代表手势采集阶段IMU数据的对象，包含两个子字段： - frequency（采样频率）：IMU传感器数据的采样频率（单位：赫兹，Hz），所有文件中该值均为200； - data（IMU数据数组）：为长度400的对象数组。每个对象包含三个字段：陀螺仪（由3个浮点值组成的三轴数组）、加速度（由3个浮点值组成的三轴加速度数组）以及旋转向量（由4个浮点值组成的四元数数组）。 本数据集可用于训练与测试自动手势识别、尤其是自动手语解读相关的算法与技术。