SharoonArshad/how2sign-processed-common-450

How2Sign Processed Common 450 Features

This repository contains the processed How2Sign dataset as a ZIP archive.

File

how2sign_processed_common_450.zip

Feature Format

Original How2Sign holistic keypoints:

(T, 543, 3)

Processed format:

(T, 450)

Where:

75 landmarks = 33 pose + 21 left hand + 21 right hand
225 position features = 75 * 3
225 velocity features = frame-to-frame difference
450 final features = position + velocity

Important

The sequence length T is variable.

Do not resize How2Sign samples to 200 frames for CTC training.

How to Use in Kaggle

from huggingface_hub import hf_hub_download
import zipfile
from pathlib import Path

repo_id = "SharoonArshad/how2sign-processed-common-450"

zip_path = hf_hub_download(
    repo_id=repo_id,
    repo_type="dataset",
    filename="how2sign_processed_common_450.zip"
)

extract_dir = Path("/kaggle/working/how2sign_processed_common_450_unzipped")
extract_dir.mkdir(parents=True, exist_ok=True)

with zipfile.ZipFile(zip_path, "r") as zip_ref:
    zip_ref.extractall(extract_dir)

print("Extracted to:", extract_dir)

After extraction, the dataset folder will contain:

how2sign_processed_common_450/
├── features/
├── metadata/
├── reports/
└── README.md

Processing Stats

{
  "train": {
    "num_samples": 31047,
    "min_frames": 1,
    "max_frames": 2579,
    "mean_frames": 162.75624053853835,
    "median_frames": 126.0,
    "metadata_matched": 0,
    "target_text_available": 0
  },
  "val": {
    "num_samples": 1739,
    "min_frames": 3,
    "max_frames": 1761,
    "mean_frames": 158.2708453133985,
    "median_frames": 124.0,
    "metadata_matched": 0,
    "target_text_available": 0
  },
  "test": {
    "num_samples": 2343,
    "min_frames": 1,
    "max_frames": 1319,
    "mean_frames": 162.78787878787878,
    "median_frames": 127.0,
    "metadata_matched": 0,
    "target_text_available": 0
  }
}