Dom/rpa_vision_v3
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: template matching sur fenêtre active + seuil 0.90

Browse Source 
Template matching des icônes limité à la fenêtre active (window.png)
pour éviter les faux positifs sur le full screen. Seuil relevé de
0.70 à 0.90. Coordonnées fenêtre converties en coordonnées écran.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Dom
2026-04-04 23:23:21 +02:00
parent 84a91630e9
commit 90ee8ca8f4
1 changed files with 22 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
agent_v0/server_v1/api_stream.py
View File

@@ -5200,14 +5200,31 @@ def _resolve_target_sync(
return grounding_result return grounding_result
if not by_text_strict or by_text_source != "ocr": if not by_text_strict or by_text_source != "ocr":
# Template matching sur la fenêtre active si disponible (évite les faux positifs)
window_capture = target_spec.get("window_capture", {})
window_rect = window_capture.get("rect")
from pathlib import Path as _Path
_full = _Path(screenshot_path)
_win = _full.parent / _full.name.replace("_full.png", "_window.png")
tm_path = str(_win) if _win.is_file() and window_rect else screenshot_path
tm_screen_w = (window_rect[2] - window_rect[0]) if window_rect and _win.is_file() else screen_width
tm_screen_h = (window_rect[3] - window_rect[1]) if window_rect and _win.is_file() else screen_height
result = _resolve_by_template_matching( result = _resolve_by_template_matching(
screenshot_path=screenshot_path, screenshot_path=tm_path,
anchor_image_b64=anchor_image_b64, anchor_image_b64=anchor_image_b64,
screen_width=screen_width, screen_width=tm_screen_w,
screen_height=screen_height, screen_height=tm_screen_h,
confidence_threshold=0.70, confidence_threshold=0.90,
) )
if result and result.get("score", 0) >= 0.70: if result and result.get("score", 0) >= 0.90:
x_tm, y_tm = result["x_pct"], result["y_pct"]
# Convertir coordonnées fenêtre → écran si nécessaire
if window_rect and _win.is_file():
abs_x = window_rect[0] + x_tm * tm_screen_w
abs_y = window_rect[1] + y_tm * tm_screen_h
result["x_pct"] = round(abs_x / screen_width, 6)
result["y_pct"] = round(abs_y / screen_height, 6)
logger.info( logger.info(
"Strict resolve TEMPLATE : icon match (score=%.3f)", "Strict resolve TEMPLATE : icon match (score=%.3f)",
result.get("score", 0), result.get("score", 0),