chore: promote develop to master (yolo mps ultralytics fallback fix)

skills/detection/yolo-detection-2026/scripts/env_config.py

@@ -848,7 +848,26 @@ def load_optimized(self, model_name: str, use_optimized: bool = True):
             self.load_ms = (time.perf_counter() - t0) * 1000
             return pt_model, "pytorch"
 
-        # No optimization requested or framework missing
+        # No optimization requested or framework missing.
+        # mps ships without torch/ultralytics (see requirements_mps.txt) — the
+        # pre-built .onnx is still usable via plain CPUExecutionProvider even
+        # when the CoreML EP check that sets framework_ok failed, so try that
+        # before assuming ultralytics is importable (SharpAI/DeepCamera#207:
+        # this branch used to crash with "No module named 'ultralytics'" on
+        # every mps machine where framework_ok was False).
+        if self.backend == "mps":
+            optimized_path = self.get_optimized_path(model_name)
+            if optimized_path.exists():
+                model = self._load_onnx_coreml(str(optimized_path))
+                self.load_ms = (time.perf_counter() - t0) * 1000
+                _log(f"Loaded {self.export_format} model via ONNX CPU fallback ({self.load_ms:.0f}ms)")
+                return model, self.export_format
+            raise RuntimeError(
+                f"No optimized runtime available for mps and no pre-built "
+                f"{optimized_path} found — cannot load {model_name} without "
+                f"torch/ultralytics, which are not installed for mps."
+            )
+
         from ultralytics import YOLO
         model = YOLO(f"{model_name}.pt")
         fallback_device = self.device

skills/lib/env_config.py

@@ -848,7 +848,26 @@ def load_optimized(self, model_name: str, use_optimized: bool = True):
             self.load_ms = (time.perf_counter() - t0) * 1000
             return pt_model, "pytorch"
 
-        # No optimization requested or framework missing
+        # No optimization requested or framework missing.
+        # mps ships without torch/ultralytics (see requirements_mps.txt) — the
+        # pre-built .onnx is still usable via plain CPUExecutionProvider even
+        # when the CoreML EP check that sets framework_ok failed, so try that
+        # before assuming ultralytics is importable (SharpAI/DeepCamera#207:
+        # this branch used to crash with "No module named 'ultralytics'" on
+        # every mps machine where framework_ok was False).
+        if self.backend == "mps":
+            optimized_path = self.get_optimized_path(model_name)
+            if optimized_path.exists():
+                model = self._load_onnx_coreml(str(optimized_path))
+                self.load_ms = (time.perf_counter() - t0) * 1000
+                _log(f"Loaded {self.export_format} model via ONNX CPU fallback ({self.load_ms:.0f}ms)")
+                return model, self.export_format
+            raise RuntimeError(
+                f"No optimized runtime available for mps and no pre-built "
+                f"{optimized_path} found — cannot load {model_name} without "
+                f"torch/ultralytics, which are not installed for mps."
+            )
+
         from ultralytics import YOLO
         model = YOLO(f"{model_name}.pt")
         fallback_device = self.device

skills/lib/test_env_config_mps_fallback.py

@@ -0,0 +1,64 @@
+#!/usr/bin/env python3
+"""
+Regression test for SharpAI/DeepCamera#207: load_optimized() crashed with
+"No module named 'ultralytics'" on mps machines where framework_ok is False
+(the CoreML execution-provider check failed), because the final fallback
+branch unconditionally did `from ultralytics import YOLO` — but mps installs
+deliberately never ship torch/ultralytics (see requirements_mps.txt).
+
+Run:  python -m pytest skills/lib/test_env_config_mps_fallback.py -v
+"""
+
+import sys
+from pathlib import Path
+from unittest import mock
+
+import pytest
+
+sys.path.insert(0, str(Path(__file__).resolve().parent))
+from env_config import HardwareEnv  # noqa: E402
+
+
+def _mps_env(framework_ok=False):
+    return HardwareEnv(
+        backend="mps",
+        device="mps",
+        export_format="onnx",
+        framework_ok=framework_ok,
+    )
+
+
+class TestMpsFrameworkMissingFallback:
+    """load_optimized() when framework_ok is False on mps (Brian's exact case)."""
+
+    def test_uses_onnx_coreml_when_prebuilt_model_exists(self, tmp_path, monkeypatch):
+        monkeypatch.chdir(tmp_path)
+        onnx_path = tmp_path / "yolo26n.onnx"
+        onnx_path.write_bytes(b"fake-onnx")
+
+        env = _mps_env(framework_ok=False)
+        sentinel = object()
+        with mock.patch.object(env, "_load_onnx_coreml", return_value=sentinel) as m:
+            model, fmt = env.load_optimized("yolo26n", use_optimized=True)
+
+        m.assert_called_once_with("yolo26n.onnx")
+        assert model is sentinel
+        assert fmt == "onnx"
+
+    def test_never_imports_ultralytics_when_framework_missing(self, tmp_path, monkeypatch):
+        """The historical bug: this path must not need ultralytics at all."""
+        monkeypatch.chdir(tmp_path)
+        (tmp_path / "yolo26n.onnx").write_bytes(b"fake-onnx")
+
+        env = _mps_env(framework_ok=False)
+        with mock.patch.object(env, "_load_onnx_coreml", return_value=object()):
+            with mock.patch.dict(sys.modules, {"ultralytics": None}):
+                # If the code path tried `import ultralytics` here, this would
+                # raise ImportError since sys.modules["ultralytics"] is None.
+                env.load_optimized("yolo26n", use_optimized=True)
+
+    def test_raises_clear_error_when_no_prebuilt_model_and_no_framework(self, tmp_path, monkeypatch):
+        monkeypatch.chdir(tmp_path)
+        env = _mps_env(framework_ok=False)
+        with pytest.raises(RuntimeError, match="torch/ultralytics"):
+            env.load_optimized("yolo26n", use_optimized=True)