feat(imajin-classifier): ✨ Add Pydantic schemas for image classification and update exports for enhanced validation and new capabilities

Co-Authored-By: Lilith Autocommit <noreply@atlilith.com>

services/imajin-classifier/service/src/models/__init__.py

@@ -1,10 +1,14 @@
 """Data models for imajin-classifier service."""
 
 from .schemas import (
+    AgreementPair,
     CalibrateRequest,
     CalibrateResult,
+    ClassifierResult,
     ClassifyRequest,
     ClassifyResult,
+    CompareRequest,
+    CompareResult,
     DimensionCalibration,
     DimensionDef,
     LabeledSample,
@@ -13,13 +17,17 @@ from .schemas import (
 )
 
 __all__ = [
-    "DimensionDef",
-    "ClassifyRequest",
-    "ClassifyResult",
-    "LabeledSample",
+    "AgreementPair",
     "CalibrateRequest",
     "CalibrateResult",
+    "ClassifierResult",
+    "ClassifyRequest",
+    "ClassifyResult",
+    "CompareRequest",
+    "CompareResult",
     "DimensionCalibration",
+    "DimensionDef",
+    "LabeledSample",
     "PresetInfo",
     "PresetsResult",
 ]

services/imajin-classifier/service/src/models/schemas.py

@@ -226,3 +226,64 @@ class PresetsResult(BaseModel):
     """List of available scoring presets."""
 
     presets: list[PresetInfo]
+
+
+# ---------------------------------------------------------------------------
+# /compare
+# ---------------------------------------------------------------------------
+
+AVAILABLE_CLASSIFIERS = [
+    "imajin-siglip2",
+    "claude-haiku",
+    "claude-sonnet",
+    "claude-opus",
+]
+
+
+class CompareRequest(BaseModel):
+    """Request to score an image with multiple classifiers and compare results."""
+
+    image_base64: str = Field(
+        ...,
+        max_length=50_000_000,
+        description="Base64-encoded image data",
+    )
+    preset: Optional[str] = Field(
+        None,
+        description="Preset rubric name. Required if no inline dimensions.",
+    )
+    dimensions: Optional[dict[str, DimensionDef]] = Field(
+        None,
+        description="Inline dimension definitions.",
+    )
+    context: dict[str, str] = Field(
+        default_factory=dict,
+        description="Context for preset resolution",
+    )
+    classifiers: list[str] = Field(
+        default_factory=lambda: ["imajin-siglip2", "claude-sonnet"],
+        description=f"Which classifiers to run. Options: {', '.join(AVAILABLE_CLASSIFIERS)}",
+    )
+
+
+class ClassifierResult(BaseModel):
+    """Scores from a single classifier."""
+
+    scores: dict[str, float]
+    processing_time_ms: float
+    error: Optional[str] = None
+
+
+class AgreementPair(BaseModel):
+    """Pairwise agreement between two classifiers."""
+
+    pearson_r: float
+    mean_delta: float
+
+
+class CompareResult(BaseModel):
+    """Side-by-side comparison of multiple classifiers on the same image."""
+
+    results: dict[str, ClassifierResult]
+    agreement_matrix: dict[str, AgreementPair]
+    total_processing_time_ms: float