feat: initial HSAP platform

Huaxu Sentinel Active Safety Platform with embedded algorithm code,
Docker Compose setup, and vendored dataset scaffolds for clone-and-run.

Co-authored-by: Cursor <cursoragent@cursor.com>

algorithms/lane_ufld/adapter.py

@@ -0,0 +1,108 @@
+"""Lane UFLD 引擎适配。"""
+from __future__ import annotations
+
+import subprocess
+import sys
+from pathlib import Path
+from typing import Any
+
+import yaml
+
+WORKSPACE = Path(__file__).resolve().parents[2]
+
+
+def _lane_workdir() -> Path:
+    wf = yaml.safe_load((WORKSPACE / "workflow.registry.yaml").read_text(encoding="utf-8"))
+    return (WORKSPACE / wf["projects"]["lane"]["train"]["workdir"]).resolve()
+
+
+def _latest_lane_weights(workdir: Path) -> str | None:
+    candidates = sorted(workdir.rglob("best.pth"), key=lambda p: p.stat().st_mtime, reverse=True)
+    if not candidates:
+        return None
+    return str(candidates[0].resolve())
+
+
+def train_local(config_overrides: dict | None = None) -> dict[str, Any]:
+    wd = _lane_workdir()
+    cfg = yaml.safe_load((WORKSPACE / "workflow.registry.yaml").read_text())["projects"]["lane"]["train"]["config"]
+    proc = subprocess.run(
+        [sys.executable, "train.py", cfg],
+        cwd=str(wd),
+        capture_output=True,
+        text=True,
+    )
+    if proc.returncode != 0:
+        raise RuntimeError(proc.stderr or proc.stdout)
+    return {
+        "ok": True,
+        "stdout": proc.stdout,
+        "stderr": proc.stderr,
+        "track": "local",
+        "workdir": str(wd),
+        "command": f"{sys.executable} train.py {cfg}",
+        "best_weights": _latest_lane_weights(wd),
+    }
+
+
+def train_platform() -> dict[str, Any]:
+    wd = _lane_workdir()
+    wf = yaml.safe_load((WORKSPACE / "workflow.registry.yaml").read_text(encoding="utf-8"))
+    cfg = wf["projects"]["lane"]["train"]["config"]
+    proc = subprocess.run([sys.executable, "train.py", cfg], cwd=str(wd), capture_output=True, text=True)
+    if proc.returncode != 0:
+        raise RuntimeError(proc.stderr or proc.stdout)
+    return {
+        "ok": True,
+        "stdout": proc.stdout,
+        "stderr": proc.stderr,
+        "track": "platform",
+        "workdir": str(wd),
+        "command": f"{sys.executable} train.py {cfg}",
+        "best_weights": _latest_lane_weights(wd),
+    }
+
+
+def eval_task(model_path: str | None = None, data_root: str | None = None, test_list: str = "list/test_gt.txt") -> dict[str, Any]:
+    if not model_path:
+        raise ValueError("lane eval 需要 model_path（best.pth）")
+    wd = _lane_workdir()
+    wf = yaml.safe_load((WORKSPACE / "workflow.registry.yaml").read_text(encoding="utf-8"))
+    cfg = wf["projects"]["lane"]["train"]["config"]
+    root = data_root or str((WORKSPACE / wf["projects"]["lane"]["root"]).resolve())
+    model = model_path
+    cmd = [sys.executable, "test.py", cfg, "--test_model", model, "--data_root", root, "--test_list", test_list]
+    proc = subprocess.run(cmd, cwd=str(wd), capture_output=True, text=True)
+    if proc.returncode != 0:
+        raise RuntimeError(proc.stderr or proc.stdout)
+    return {
+        "ok": True,
+        "stdout": proc.stdout,
+        "stderr": proc.stderr,
+        "workdir": str(wd),
+        "command": " ".join(cmd),
+        "model_path": model,
+    }
+
+
+def visualize_task(model_path: str | None = None, data_root: str | None = None, test_list: str = "list/test_gt.txt") -> dict[str, Any]:
+    if not model_path:
+        raise ValueError("lane visualize 需要 model_path（best.pth）")
+    wd = _lane_workdir()
+    wf = yaml.safe_load((WORKSPACE / "workflow.registry.yaml").read_text(encoding="utf-8"))
+    cfg = wf["projects"]["lane"]["train"]["config"]
+    root = data_root or str((WORKSPACE / wf["projects"]["lane"]["root"]).resolve())
+    model = model_path
+    cmd = [sys.executable, "demo.py", cfg, "--test_model", model, "--data_root", root]
+    proc = subprocess.run(cmd, cwd=str(wd), capture_output=True, text=True)
+    if proc.returncode != 0:
+        raise RuntimeError(proc.stderr or proc.stdout)
+    return {
+        "ok": True,
+        "stdout": proc.stdout,
+        "stderr": proc.stderr,
+        "workdir": str(wd),
+        "command": " ".join(cmd),
+        "model_path": model,
+        "note": f"建议测试列表: {test_list}",
+    }