tngo/product-recognition-demo
1
0
Fork 0
forked from vnguyen/product-recognition-demo
Code Pull requests Activity

fix(recognition): calibrate the accept lines against unrelated shelves

Browse source 
Every threshold was fitted on image6, where each crop is a facing of a product
the shelf holds, so the set could not price a false accept. Measured on 4580
crops from 64 unrelated Wikimedia shelf photos at the shipped configuration:

  LOCAL_ACCEPT 2.0 -> 4.0   false accepts 60 (1.31%) -> 3 (0.07%)
  TEXT_MIN_SHARED 2 -> 4    the two-bigram reads that named sku_l12 via おい/いし

On image6 that is 44 shown / 42 correct at full resolution - 95.5% precision,
65.6% coverage - the highest-coverage cell of the joint sweep that passes both
G2 (>= 95%) and G3 (<= 1%). At live resolution 56% end-to-end at 97%.

The text lane's evidence floor gates its answer only: zeroing the row's score
would reorder the ranking and make the margin gate compare a different pair
than the one it was measured on.

code/sweep_accept_lines.py sweeps both axes jointly through the shipped
DecisionPolicy; the earlier table swept one axis while pinning the other, so
its baseline rows were not the old line.
This commit is contained in:
kiyo9w 2026-09-29 09:28:13 +09:00
parent 4ce41ecb17
commit 5094b5f6ca
12 changed files with 815 additions and 195 deletions
Download patch file Download diff file Expand all files Collapse all files

29
recognization/code/test_live.py
View file

@ -455,6 +455,35 @@ class LocalScaleTests(unittest.TestCase):
self.assertGreater(engine.ocr_calls, 1, "the size-only read must not be cached")
engine.shutdown()
def test_a_failing_lane_is_counted_and_logged(self):
"""A lane that raises is a failure, not a crop that matched nothing.
`_on_classified` substituted an empty answer for the exception, so a
run in which `score_crop_lanes` failed on every crop produced tracks,
zero labels and a clean log - indistinguishable from a real coverage
drop. The counter and the log line are what tell the two apart, and
the session has to keep serving frames either way.
"""
session, engine, _, _ = make_session(engine_cls=LocalScaleEngine)
def boom(crop, method=None, skip_ocr=False):
raise RuntimeError("lane unavailable")
engine.score_crop_lanes = boom
named = None
with self.assertLogs("dorayaki.live", level="ERROR") as logs:
for frame_id in range(1, 14):
tracks = process(session, frame_id)["tracks"]
self.assertTrue(tracks, "a failing lane must not take the session down")
named = named or next(
(t["identity"]["sku_id"] for t in tracks if t["identity"]["sku_id"]), None)
self.assertTrue(wait_for(lambda: bool(logs.output)),
"the exception must be logged, not swallowed")
self.assertGreaterEqual(session.snapshot()["stats"]["classify_failures"], 1,
"a lane failure must be counted in the session stats")
self.assertIsNone(named, "a failed lane must not name a product")
engine.shutdown()
class TagPassIntegrationTests(unittest.TestCase):
"""The tag pass must reach the live result without blocking frames."""