How blzs-zhjk/dynmultidep works
This is a small LMVD visual-feature conversion utility. When imported or run, extract_npy_lmvd.py walks ./lmvd/Video_feature, processes CSV files not already converted, selects facial landmark coordinates, keeps every 30th frame, scales each row, and saves NumPy arrays under ./lmvd/visual. The README is empty, so the workflow description is based on the sampled script; the input and output locations are hard-coded there.
Main entry points: extract_npy_lmvd.py Batch traversal
Components 5
Other components
- Python runtime
- Batch traversal
extract_npy_lmvd.py - Landmark transform
extract_npy_lmvd.py - LMVD CSV files
- Visual NPY files
Connections 4
- Python runtime → Batch traversal: runs at import
- Batch traversal → LMVD CSV files: scans and reads
- Batch traversal → Landmark transform: processes new files
- Landmark transform → Visual NPY files: saves arrays