Table 3 Dataset naming conventions and structures.

From: A Structural Vibration-based Dataset for Human Gait Recognition

Dataset

Data Type

Structure

Naming Convention

 

Raw

Folders P1-P100, each with ≥4 .mat files

PX_Y.mat (e.g., P1_1.mat)

A1

Interim

Single .mat file

A1.mat (1501 columns: features + person_id 1-100)

 

Processed

Folders P1-P100 with images

PX/event_Z.jpg (224 × 224 CWT images)

 

Raw

A2_* (1.5m/2.5m/4.0m)  →  P1-P30

PX_Y.mat

A2

Interim

Three .mat files

A2_*.mat (person_id 1-30 per distance)

 

Processed

A2_1, A2_2, A2_3 →  P1-P30

Subfolder per distance, then participant

 

Raw

A3_* (Wooden/Carpet/Cement)  →  P1-P40

PX_Y.mat

A3

Interim

Three .mat files

A3_*.mat (person_id 1-40 per floor type)

 

Processed

A3_1, A3_2, A3_3 →  P1-P40

Subfolder per floor, then participant

 

Raw

P1-P15 folders

PX_1.mat (one per participant)

A4

Interim

A4_1.mat + image folders

A4_1.mat, A4_2a/PX.jpg, A4_2b/PX.jpg

 

Processed

A4_1 →  P1-P15

Participant folders with CWT images

 

Raw

A5_* (speed)  →  P1-P15

PX_Y.mat (≥2 files/participant)

A5

Interim

Three .mat files

A5_*.mat (person_id 1-15 per speed)

 

Processed

A5_1, A5_2, A5_3 →  P1-P15

Subfolder per speed, then participant

  1. Each dataset includes raw, interim, and processed stages, with specific folder and file naming schemes.