fix(perception): disable PaddleOCR doc-unwarping for screenshots

PaddleOCR's OCR.yaml pipeline defaults to use_doc_orientation_classify
and use_doc_unwarping enabled, which are meant for photographed paper
documents. Applied to a flat, upright device screenshot, UVDoc
geometrically warps the image before detection, and returns box
coordinates in that warped space with no inverse mapping back to the
original image.

Verified on a real screenshot: with unwarping on, the same detected
element ("新项目") shifts from y=158 to y=71 versus the original image,
and 2 boxes near the top edge (status bar time/battery) are dropped
entirely. Disabling both flags by default (still overridable via
explicit kwargs) makes detected boxes match the original screenshot.

perception/ocr.py

@@ -33,6 +33,16 @@ class OCRBox:
 
 class PaddleOCREngine:
     def __init__(self, **kwargs: Any) -> None:
+        # Device screenshots are already upright, flat digital captures, not
+        # photographed paper documents, so PaddleOCR's document-preprocessing
+        # models (orientation classification + UVDoc unwarping) have nothing
+        # real to correct. Left at their library defaults (True), they still
+        # run, geometrically warp the image, and detect/recognize text against
+        # that warped image, returning box coordinates that no longer line up
+        # with the original screenshot. Callers can still opt back in via an
+        # explicit kwarg.
+        kwargs.setdefault("use_doc_orientation_classify", False)
+        kwargs.setdefault("use_doc_unwarping", False)
         self.kwargs = kwargs
         self._engine: Any | None = None
 

tests/test_ocr.py

@@ -4,7 +4,24 @@ import numpy as np
 import pytest
 
 from core.models import Bounds
-from perception.ocr import OCRBox, parse_paddle_result, run_ocr
+from perception.ocr import OCRBox, PaddleOCREngine, parse_paddle_result, run_ocr
+
+
+def test_paddle_ocr_engine_disables_doc_preprocessing_by_default() -> None:
+    # Screenshots are flat, upright digital captures, not photographed paper
+    # documents: PaddleOCR's document-unwarping preprocessing has nothing real
+    # to correct and instead warps the image, so detected box coordinates no
+    # longer line up with the original screenshot.
+    engine = PaddleOCREngine(lang="ch")
+
+    assert engine.kwargs["use_doc_orientation_classify"] is False
+    assert engine.kwargs["use_doc_unwarping"] is False
+
+
+def test_paddle_ocr_engine_lets_callers_override_doc_preprocessing() -> None:
+    engine = PaddleOCREngine(use_doc_unwarping=True)
+
+    assert engine.kwargs["use_doc_unwarping"] is True
 
 
 def test_parse_paddle_result_accepts_ndarray_fields() -> None: