Dharunkumar9/Fingerspelling_dataset

SHWASI Flex Sensor Fingerspelling Dataset

This dataset contains wearable 5-finger flex sensor recordings for fingerspelling/sign combinations, published with the SHWASI work.

What is in this repository

• Format: .xlsx files (Excel), not CSV
• Total files: 96
• Folder structure:
  • MODE 1/ (32 files)
  • MODE 2/ (32 files)
  • MODE 3/ (32 files)

Each file corresponds to one encoded class/gesture entry in that mode. File names include the mode and encoded class (for example, Mode 1 - 00001 - Good Morning.xlsx).

Sensor setup

• 5 flex sensors: thumb, index, middle, ring, little
• ADC readings from Arduino-based acquisition setup

Load all files with metadata (Python)

from pathlib import Path
import pandas as pd

root = Path(".")  # dataset root after download
xlsx_files = sorted(root.glob("MODE */*.xlsx"))

frames = []
for fp in xlsx_files:
    df = pd.read_excel(fp)

    # Example filename:
    # Mode 1 - 00001 - Good Morning.xlsx
    stem = fp.stem
    parts = [p.strip() for p in stem.split("-")]

    mode = fp.parent.name  # MODE 1 / MODE 2 / MODE 3
    class_code = parts[1] if len(parts) > 1 else None
    label_text = "-".join(parts[2:]).strip() if len(parts) > 2 else None

    df["mode"] = mode
    df["class_code"] = class_code
    df["label_text"] = label_text
    df["source_file"] = fp.name

    frames.append(df)

full_df = pd.concat(frames, ignore_index=True)
print(full_df.shape)
print(full_df.head())

Recommended citation

@INPROCEEDINGS{10958541,
  author={E, Mithun and S, Dharun Kumar and A, Abirami and J, Mathiarasun R and E, Harikumar M},
  booktitle={2024 IEEE 21st India Council International Conference (INDICON)},
  title={(SHWASI) Smart Hand Wearable Aid for Speech Impaired: Sign Language Communication using Flex Sensor-based Finger Spelling},
  year={2024},
  pages={1-6},
  doi={10.1109/INDICON63790.2024.10958541}
}

License

This dataset is released under CC BY 4.0. Please provide attribution when using it in research or products.