docs(wiki): Albanian-plate OCR benchmark — keep cct-xs-v2-global default

Benchmarked fast-alpr's four fast-plate-ocr models via the full pipeline on real AL
plates (AA558EE, AA687KE), CPU. All four read both correctly; the default
cct-xs-v2-global-model wins on confidence (0.999/1.000) AND speed (33-39ms) and returns
region=Albania. The "European 40+country" model is WORSE here (~0.77 confidence, one
synthetic misread) — overturning the "EU model better for AL" assumption from the prior
research. Decision: no config change. Resolves the AL-accuracy-benchmark open item
(results table + finding added to opencv-anpr-service); weight-provenance remains the
one open recognizer item. Re-benchmark on real on-site captures once cameras installed.

Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V

wiki/entities/opencv-anpr-service.md

@@ -109,7 +109,27 @@ disappoints).
 
 **Recommendation:** prototype with **fast-alpr** now (permissive, offline, ONNX, fits the decided
 shape); plan a YOLO-detector fine-tune + PaddleOCR only if production accuracy demands it. Choice kept
-**open** pending the weight-provenance check + an accuracy benchmark on real AL plates.
+**open** pending the weight-provenance check (the AL-plate benchmark below is now done).
+
+### Albanian-plate OCR benchmark — keep the default (2026-06-19)
+
+Ran the four candidate `fast-plate-ocr` models through the **full pipeline** (YOLOv9 detect → OCR) on
+real AL plate photos (Wikimedia), CPU, scaffolded service:
+
+| OCR model | `AA 558 EE` | `AA 687 KE` | Speed | Note |
+| --- | --- | --- | --- | --- |
+| **`cct-xs-v2-global-model`** (default) | ✓ 0.999 | ✓ **1.000** | **33–39 ms** | best accuracy + fastest; returns `region=Albania` |
+| `cct-s-v2-global-model` | ✓ 0.998 | ✓ 0.999 | 50–65 ms | as accurate, ~50% slower |
+| `global-plates-mobile-vit-v2-model` | ✓ 0.955 | ✓ 0.959 | 33–35 ms | fast, lower confidence |
+| `european-plates-mobile-vit-v2-model` | ✓ 0.784 | ✓ 0.766 | 38–46 ms | correct but **much lower confidence**; misread a synthetic `AB123FG`→`AB123FO` |
+
+**Finding (overturns the prior assumption):** the **default `cct-xs-v2-global-model` is the best for
+Albania** — most accurate AND fastest. The "European (40+ country)" model is *worse* here (~0.77 vs
+~1.0 confidence, one synthetic misread), despite the "EU model → better for AL" intuition. So **no
+config change**: `VISION_OCR_MODEL` stays `cct-xs-v2-global-model`. Caveat: both test photos were
+clean head-on shots; real booth captures (angled, dirty, night, motion-blur) will lower absolute
+confidence — the `min_confidence=0.5` floor (→ `low_confidence` → ticket-path fallback) covers that.
+The ranking should hold; re-benchmark on real on-site captures once the cameras are installed.
 
 ## Anti-fraud / threat-model fit
 
@@ -125,9 +145,10 @@ shape); plan a YOLO-detector fine-tune + PaddleOCR only if production accuracy d
 
 ## Open
 
-- **Recognizer choice** — **fast-alpr (MIT, YOLOv9+CCT on ONNX) is the evaluated baseline** (see the
-  Recognizer evaluation section above); remaining open items are the **weight-provenance check** and
-  an **accuracy benchmark on real AL plates** (default global vs. the 40+country EU model). See
+- **Recognizer choice** — **fast-alpr (MIT, YOLOv9+CCT on ONNX) is the baseline, AL-benchmarked**: the
+  default `cct-xs-v2-global-model` won over the EU model on real AL plates (table above). The one
+  remaining open item is the **model-weight-provenance check** (the MIT-weights claim). A re-benchmark
+  on real *on-site* captures (angled/night/dirty) is wanted once cameras are installed. See
   [[vision-service]]; AGPL still permitted in-service for the stronger fallback.
 - **Vehicle fingerprint**: attribute classifier vs. embedding-similarity; what threshold makes a
   mismatch an anomaly without false-positiving on lighting/angle.