text stringlengths 0 86 |
|---|
This dataset includes activity data from 51 participants from an activity |
recognition project. Each participant performed each of the 18 activities |
(listed in activity_key.txt) for 3 minutes and the sensor data (accelerometer |
and gyroscope for smartphone and smartwatch) was recorded at a rate of 20 Hz. |
The smartphones used were Nexus 5, Nexus 5X, and Galaxy S6 while the smartwatch |
used was the LG G watch. This dataset contains the raw data as well as the |
transformed data which was generated using arffmagic (located in |
arffmagic-master). |
A much more detailed description of the two main data sets is provided in the |
wisdm-dataset-description.pdf document at the top level of the data directory. |
################################################################################ |
################################################################################ |
The raw sensor data is located in the raw directory. Each user has its own data |
file which is tagged with their subject id, the sensor, and the device. Within |
the data file, each line is: |
Subject-id, Activity Label, Timestamp, x, y, z |
The features are defined as follows: |
subject-id: Identfies the subject and is an integer value between 1600 and 1650. |
activity-label: see activity_key.txt for a mapping from 18 characters to the |
activity name |
timestamp: time that the reading was taken (Unix Time) |
x: x sensor value (real valued) |
y: y sensor value (real valued) |
z: z sensor value (real valued) |
################################################################################ |
################################################################################ |
ARFF files for each user (with attributes described below) are located in the |
arff_files directory. The ARFF files are created using the arffmagic program, which |
aggregates the lower level raw data into examples labeled with the activity being |
performed. The descriptions of these features are also found in the many WISDM |
activity recognition papers. |
Recognize that the sensors are samples at 20Hz and we use a 10 second window size, |
so most of the statistics are based on the aggregation of 200 sensor readings. |
Attribute information: |
ACTIVITY[1]: |
This field contains the code that uniquely identifies the activity |
(activity_key.txt provides the mapping from code to actual activity) |
Binned Distribution[30]: |
The range of values is determined (maximum - minimum), 10 equal-sized |
bins are formed, and the fraction of the 200 values within each bin |
is recorded for each of the 3 axes. The axis bin values are provided |
in the following order: x, y, z (10 values for each). The attributes are |
named X0 .. X9, Y1 ... Y9, and Z0 ... Z9 |
Average[3]: |
Average sensor value (for each of the 3 axes). Labeled XAVG, YAVG, ZAVG |
Time Between Peaks[3]: |
Time between peaks in the sinusoidal waves formed by the data |
as determined by a simple algorithm (feel free to check the code). Done |
for each axis). Labeled XPEAK, YPEAK, ZPEAK. |
Average Absolute Difference[3]: |
Average absolute difference between the 200 values and the mean of these |
values (for each axis). Labeled as {X,Y,Z}ABSOLDEV |
Standard Deviation[3]: |
Standard deviation (for each axis). Labeled as {X,Y,Z}STANDDEV |
Variance[3]; |
The variance of the values (for each axis). Labeled as {X,Y,Z}VAR |
-------------------------------------------------------------------------------- |
THE NEXT THREE SETS OF FEATURES ARE NOT USED IN OUR PUBLISHED RESEARCH PAPERS. THEY |
WERE ADDED TO EXPERIMENT WITH. |
MFCC: Mel-frequency cepstral coefficients [39] |
MFCCs are a representation of the short-term power spectrum of a wave, based |
on a linear cosine transform of a log power spectrum on a nonlinear mel scale |
of frequency. There are 13 per axis There are 13 per axis. They are labeled as |
XMFCC{0-12}, YMFCC{0-12}, and ZMFCC{0-12}. |
Cosine distance [3]: |
These are the cosine distances between the sensor values for a pair of axes. |
Three pairs are considered to cover all possible pairs. They are labeled as |
{XY, XZ, YZ}COS. |
Correlation [3]: |
These are the correlations between the sensor values for a pair of axes. |
Three pairs are considered to cover all possible pairs. They are labeled as |
{XY, XZ, YZ}COR. |
---------------------------------------------------------------------------------- |
End of preview. Expand
in Data Studio
YAML Metadata Warning: The task_categories "time-series-classification" is not in the official list: text-classification, token-classification, table-question-answering, question-answering, zero-shot-classification, translation, summarization, feature-extraction, text-generation, fill-mask, sentence-similarity, text-to-speech, text-to-audio, automatic-speech-recognition, audio-to-audio, audio-classification, audio-text-to-text, voice-activity-detection, depth-estimation, image-classification, object-detection, image-segmentation, text-to-image, image-to-text, image-to-image, image-to-video, unconditional-image-generation, video-classification, reinforcement-learning, robotics, tabular-classification, tabular-regression, tabular-to-text, table-to-text, multiple-choice, text-ranking, text-retrieval, time-series-forecasting, text-to-video, image-text-to-text, image-text-to-image, image-text-to-video, visual-question-answering, document-question-answering, zero-shot-image-classification, graph-ml, mask-generation, zero-shot-object-detection, text-to-3d, image-to-3d, image-feature-extraction, video-text-to-text, keypoint-detection, visual-document-retrieval, any-to-any, video-to-video, other
gem-analysis-wisdm
WISDM activity prediction dataset
数据集信息
- 来源路径:
datasets/wisdm - 数据大小: 895.2 MB
- 用途: 健身动作识别模型训练
使用方法
from huggingface_hub import snapshot_download
# 下载数据集
snapshot_download(
repo_id="yonful/gem-analysis-wisdm",
repo_type="dataset",
local_dir="./datasets/wisdm"
)
或使用项目中的下载脚本:
python scripts/prepare_data.py --dataset wisdm
许可证
请参考原始数据源的许可证要求。
- Downloads last month
- -