Buckets:
| """Run the data pipeline on a 500-video subset of SkatingVerse with the YOLO GPU extractor. | |
| Output goes to /data/processed_500 (isolated from the real processed/ dir). Invoked as a | |
| background job; progress is written to run_500.log. | |
| """ | |
| from __future__ import annotations | |
| import time | |
| from pathlib import Path | |
| import pipeline | |
| def main() -> int: | |
| cfg = dict(pipeline.CONFIG) | |
| cfg["datasets"] = dict(pipeline.CONFIG["datasets"]) | |
| cfg["pose_estimator"] = "yolo" | |
| cfg["yolo_weights"] = "yolo11n-pose.pt" | |
| cfg["max_videos"] = 500 | |
| cfg["output_dir"] = Path("/data/processed_500") | |
| cfg["skeleton_cache_dir"] = Path("/data/processed_500/skeleton_cache") | |
| print("=" * 72, flush=True) | |
| print("500-sample SkatingVerse run | extractor=yolo11n-pose | imgsz=640 | fp16", flush=True) | |
| print(f"output -> {cfg['output_dir']}", flush=True) | |
| print("=" * 72, flush=True) | |
| t0 = time.perf_counter() | |
| ok = pipeline.run_pipeline(cfg, sources=["skatingverse"]) | |
| print(f"\nrun_pipeline ok={ok} | wall={time.perf_counter() - t0:.1f}s", flush=True) | |
| return 0 if ok else 1 | |
| if __name__ == "__main__": | |
| raise SystemExit(main()) | |
Xet Storage Details
- Size:
- 1.15 kB
- Xet hash:
- 25a7640d0c02d38887a9b8aa0c35b1b758e87abaf37ae26bb625562b438de41d
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.