| """Run checkpoint-backed SatlasNet inference on a test NPZ.""" |
|
|
| import argparse |
| import importlib.util |
| from pathlib import Path |
| import numpy as np |
| import torch |
| import yaml |
|
|
| ROOT = Path(__file__).resolve().parents[1] |
|
|
|
|
| def main(): |
| parser = argparse.ArgumentParser(description=__doc__) |
| parser.add_argument("--config", type=Path, default=ROOT / "conf/config.yaml"); parser.add_argument("--data", type=Path) |
| parser.add_argument("--checkpoint", type=Path); parser.add_argument("--output-dir", type=Path) |
| parser.add_argument("--device", choices=("auto", "cpu", "cuda"), default="auto"); args = parser.parse_args() |
| config = yaml.safe_load(args.config.read_text()); checkpoint_path = args.checkpoint or ROOT / config["paths"]["checkpoint"] |
| if not checkpoint_path.is_file(): raise FileNotFoundError(f"checkpoint not found: {checkpoint_path}") |
| spec = importlib.util.spec_from_file_location("satlaspretrain", ROOT / "model/satlaspretrain.py") |
| module = importlib.util.module_from_spec(spec); spec.loader.exec_module(module) |
| checkpoint = torch.load(checkpoint_path, map_location="cpu", weights_only=False) |
| model = module.SatlasPretrain(**config["model"]); model.load_state_dict(checkpoint["model"]) |
| use_cuda = torch.cuda.is_available() and args.device != "cpu" |
| if args.device == "cuda" and not use_cuda: raise RuntimeError("CUDA requested but unavailable") |
| device = torch.device("cuda" if use_cuda else "cpu"); model.to(device).eval() |
| data_path = args.data or ROOT / config["data"]["root"] / "test.npz"; archive = np.load(data_path) |
| module.validate_npz(archive, config) |
| with torch.inference_mode(): |
| outputs = model(torch.from_numpy(archive["highres_images"]).to(device), |
| torch.from_numpy(archive["lowres_images"]).to(device), |
| torch.from_numpy(archive["valid_highres_times"]).to(device), |
| torch.from_numpy(archive["valid_lowres_times"]).to(device)) |
| predictions = {} |
| for name, value in outputs.items(): |
| if name in ("segmentation", "property", "classification"): value = value.softmax(1) |
| elif name != "regression": value = value.sigmoid() |
| predictions[name] = value.cpu().numpy().astype(np.float32) |
| output = args.output_dir or ROOT / config["paths"]["inference_dir"]; output.mkdir(parents=True, exist_ok=True) |
| np.savez_compressed(output / "predictions.npz", **predictions, checkpoint=np.asarray(str(checkpoint_path)), |
| sample_ids=archive["sample_ids"], |
| source=archive["source"], |
| protocol=archive["protocol"] if "protocol" in archive else np.asarray("provided_npz")) |
| print(f"inference={output / 'predictions.npz'} checkpoint={checkpoint_path}") |
|
|
|
|
| if __name__ == "__main__": main() |
|
|
