Upload README.md

README.md

@@ -0,0 +1,27 @@
+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)