Datasetpublic

meghna2801/Figure_Skating_Data

MMFS Figure Skating Dataset (Processed) This repository contains processed skeleton data for figure skating action classification. Files included: skeleton/train_data.pkl — list of numpy arrays; each sample shape (150, 17, 3, 1) skeleton/test_data.pkl — same as above for test skeleton/train_label.pkl, skeleton/test_label.pkl — integer class labels skeleton/train_name.pkl, skeleton/test_name.pkl — original action names (strings) skeleton/train_score.pkl, skeleton/test_score.pkl — action scores… See the full description on the dataset page: https://huggingface.co/datasets/meghna2801/Figure_Skating_Data.

sourceHugging Faceupdated 8mo agoView on Hugging Face

0likes2downloads

Dataset Card

MMFS Figure Skating Dataset (Processed)

This repository contains processed skeleton data for figure skating action classification.

Files included:

—skeleton/train_data.pkl — list of numpy arrays; each sample shape (150, 17, 3, 1)
—skeleton/test_data.pkl — same as above for test
—skeleton/train_label.pkl, skeleton/test_label.pkl — integer class labels
—skeleton/train_name.pkl, skeleton/test_name.pkl — original action names (strings)
—skeleton/train_score.pkl, skeleton/test_score.pkl — action scores (floats)
—train_metadata_v2.csv, test_metadata_v2.csv — metadata with data_index and cleaned action_clean
—metadata_clean.csv — compact metadata for training with data_path (if present)

How to load a sample:

python

import pickle
train_data = pickle.load(open('skeleton/train_data.pkl','rb'))
train_labels = pickle.load(open('skeleton/train_label.pkl','rb'))

i = 0
x = train_data[i]       # numpy array, shape (150,17,3,1)
y = train_labels[i]

Contact: meghna (meghna2801)