adapt junk detection

src/ui/chest.py

@@ -10,7 +10,7 @@
 
 class Chest(InventoryBase):
     def __init__(self):
-        super().__init__(5, 10)
+        super().__init__(5, 10, is_stash=True)
         self.menu_name = "Chest"
         self.is_open_search_args = SearchArgs(
             ref=["stash_menu_icon", "stash_menu_icon_medium"],

src/ui/inventory_base.py

@@ -9,7 +9,7 @@
 from utils.roi_operations import get_center, to_grid
 from utils.custom_mouse import mouse
 from utils.misc import wait
-from template_finder import get_template
+from template_finder import search
 
 
 @dataclass
@@ -26,13 +26,15 @@ class InventoryBase(Menu):
     Provides methods for identifying occupied and empty slots, item operations, etc.
     """
 
-    def __init__(self, rows: int = 3, columns: int = 11):
+    def __init__(self, rows: int = 3, columns: int = 11, is_stash: bool = False):
         super().__init__()
         self.rows = rows
         self.columns = columns
         self.slots_roi = Config().ui_roi[f"slots_{self.rows}x{self.columns}"]
-        mask_gray = cv2.cvtColor(get_template("JUNK_MASK"), cv2.COLOR_BGR2GRAY)
-        _, self.junk_mask = cv2.threshold(mask_gray, 127, 255, cv2.THRESH_BINARY)
+        if is_stash:
+            self.junk_template = "junk_stash"
+        else:
+            self.junk_template = "junk_inv"
 
     def get_item_slots(self, img: Optional[np.ndarray] = None) -> tuple[list[ItemSlot], list[ItemSlot], list[ItemSlot], list[ItemSlot]]:
         """
@@ -57,28 +59,17 @@ def get_item_slots(self, img: Optional[np.ndarray] = None) -> tuple[list[ItemSlo
             item_slot = ItemSlot(bounding_box=slot_roi, center=get_center(slot_roi))
             slot_img = crop(img, slot_roi)
 
-            junk_mask = self.junk_mask.copy()
-            if self.junk_mask.shape[:2] != slot_img.shape[:2]:
-                junk_mask = cv2.resize(junk_mask, (slot_img.shape[1], slot_img.shape[0]))
-            junk_mask_inv = cv2.bitwise_not(junk_mask)
-
             hsv_img = cv2.cvtColor(slot_img, cv2.COLOR_BGR2HSV)
             mean_value_overall = np.mean(hsv_img[:, :, 2])
             fav_flag_crop = crop(hsv_img, Config().ui_roi["rel_fav_flag"])
             mean_value_fav = cv2.mean(fav_flag_crop)[2]
 
-            # check junk
-            hsv_junk = cv2.bitwise_and(hsv_img, hsv_img, mask=junk_mask)
-            mean_hsv = cv2.mean(hsv_junk, mask=junk_mask)
-            _, mean_saturation, mean_value = mean_hsv[:3]
-            hsv_junk_inv = cv2.bitwise_and(hsv_img, hsv_img, mask=junk_mask_inv)
-            mean_hsv_inv = cv2.mean(hsv_junk_inv, mask=junk_mask_inv)
-            _, _, mean_value_inv = mean_hsv_inv[:3]
+            res_junk = search(self.junk_template, slot_img, threshold=0.7)
 
             if mean_value_fav > 205:
                 item_slot.is_fav = True
                 occupied_slots.append(item_slot)
-            elif mean_value > 110 and mean_saturation < 50 and (mean_value - mean_value_inv) > 90:
+            elif res_junk.success:
                 item_slot.is_junk = True
                 occupied_slots.append(item_slot)
             elif mean_value_overall > 37: