refactor: XYWHRect and b30 ocr

This commit is contained in:
2025-06-21 16:06:49 +08:00
parent b545c5b6bf
commit 3ebb058cdf
4 changed files with 72 additions and 60 deletions

View File

@ -1,4 +1,3 @@
from math import floor
from typing import List, Optional, Tuple
import cv2
@ -13,9 +12,21 @@ from ....ocr import (
)
from ....phash_db import ImagePhashDatabase
from ....types import Mat
from ....utils import construct_int_xywh_rect
from ...shared import B30OcrResultItem
from .colors import *
from .colors import (
BYD_MAX_HSV,
BYD_MIN_HSV,
FAR_BG_MAX_HSV,
FAR_BG_MIN_HSV,
FTR_MAX_HSV,
FTR_MIN_HSV,
LOST_BG_MAX_HSV,
LOST_BG_MIN_HSV,
PRS_MAX_HSV,
PRS_MIN_HSV,
PURE_BG_MAX_HSV,
PURE_BG_MIN_HSV,
)
from .rois import ChieriBotV4Rois
@ -25,7 +36,7 @@ class ChieriBotV4Ocr:
score_knn: cv2.ml.KNearest,
pfl_knn: cv2.ml.KNearest,
phash_db: ImagePhashDatabase,
factor: Optional[float] = 1.0,
factor: float = 1.0,
):
self.__score_knn = score_knn
self.__pfl_knn = pfl_knn
@ -72,9 +83,8 @@ class ChieriBotV4Ocr:
self.factor = img.shape[0] / 4400
def ocr_component_rating_class(self, component_bgr: Mat) -> int:
rating_class_rect = construct_int_xywh_rect(
self.rois.component_rois.rating_class_rect
)
rating_class_rect = self.rois.component_rois.rating_class_rect.rounded()
rating_class_roi = crop_xywh(component_bgr, rating_class_rect)
rating_class_roi = cv2.cvtColor(rating_class_roi, cv2.COLOR_BGR2HSV)
rating_class_masks = [
@ -89,9 +99,7 @@ class ChieriBotV4Ocr:
return max(enumerate(rating_class_results), key=lambda i: i[1])[0] + 1
def ocr_component_song_id(self, component_bgr: Mat):
jacket_rect = construct_int_xywh_rect(
self.rois.component_rois.jacket_rect, floor
)
jacket_rect = self.rois.component_rois.jacket_rect.floored()
jacket_roi = cv2.cvtColor(
crop_xywh(component_bgr, jacket_rect), cv2.COLOR_BGR2GRAY
)
@ -99,7 +107,7 @@ class ChieriBotV4Ocr:
def ocr_component_score_knn(self, component_bgr: Mat) -> int:
# sourcery skip: inline-immediately-returned-variable
score_rect = construct_int_xywh_rect(self.rois.component_rois.score_rect)
score_rect = self.rois.component_rois.score_rect.rounded()
score_roi = cv2.cvtColor(
crop_xywh(component_bgr, score_rect), cv2.COLOR_BGR2GRAY
)
@ -119,7 +127,9 @@ class ChieriBotV4Ocr:
score_roi = cv2.fillPoly(score_roi, [contour], 0)
return ocr_digits_by_contour_knn(score_roi, self.score_knn)
def find_pfl_rects(self, component_pfl_processed: Mat) -> List[List[int]]:
def find_pfl_rects(
self, component_pfl_processed: Mat
) -> List[Tuple[int, int, int, int]]:
# sourcery skip: inline-immediately-returned-variable
pfl_roi_find = cv2.morphologyEx(
component_pfl_processed,
@ -146,7 +156,7 @@ class ChieriBotV4Ocr:
return pfl_rects_adjusted
def preprocess_component_pfl(self, component_bgr: Mat) -> Mat:
pfl_rect = construct_int_xywh_rect(self.rois.component_rois.pfl_rect)
pfl_rect = self.rois.component_rois.pfl_rect.rounded()
pfl_roi = crop_xywh(component_bgr, pfl_rect)
pfl_roi_hsv = cv2.cvtColor(pfl_roi, cv2.COLOR_BGR2HSV)

View File

@ -1,12 +1,12 @@
from typing import List, Optional
from typing import List
from ....crop import crop_xywh
from ....types import Mat, XYWHRect
from ....utils import apply_factor, construct_int_xywh_rect
from ....utils import apply_factor
class ChieriBotV4ComponentRois:
def __init__(self, factor: Optional[float] = 1.0):
def __init__(self, factor: float = 1.0):
self.__factor = factor
@property
@ -19,11 +19,11 @@ class ChieriBotV4ComponentRois:
@property
def top_font_color_detect(self):
return apply_factor((35, 10, 120, 100), self.factor)
return apply_factor(XYWHRect(35, 10, 120, 100), self.factor)
@property
def bottom_font_color_detect(self):
return apply_factor((30, 125, 175, 110), self.factor)
return apply_factor(XYWHRect(30, 125, 175, 110), self.factor)
@property
def bg_point(self):
@ -31,31 +31,31 @@ class ChieriBotV4ComponentRois:
@property
def rating_class_rect(self):
return apply_factor((21, 40, 7, 20), self.factor)
return apply_factor(XYWHRect(21, 40, 7, 20), self.factor)
@property
def title_rect(self):
return apply_factor((35, 10, 430, 50), self.factor)
return apply_factor(XYWHRect(35, 10, 430, 50), self.factor)
@property
def jacket_rect(self):
return apply_factor((263, 0, 239, 239), self.factor)
return apply_factor(XYWHRect(263, 0, 239, 239), self.factor)
@property
def score_rect(self):
return apply_factor((30, 60, 270, 55), self.factor)
return apply_factor(XYWHRect(30, 60, 270, 55), self.factor)
@property
def pfl_rect(self):
return apply_factor((50, 125, 80, 100), self.factor)
return apply_factor(XYWHRect(50, 125, 80, 100), self.factor)
@property
def date_rect(self):
return apply_factor((205, 200, 225, 25), self.factor)
return apply_factor(XYWHRect(205, 200, 225, 25), self.factor)
class ChieriBotV4Rois:
def __init__(self, factor: Optional[float] = 1.0):
def __init__(self, factor: float = 1.0):
self.__factor = factor
self.__component_rois = ChieriBotV4ComponentRois(factor)
@ -100,9 +100,7 @@ class ChieriBotV4Rois:
def horizontal_items(self):
return 3
@property
def vertical_items(self):
return 10
vertical_items = 10
@property
def b33_vertical_gap(self):
@ -112,16 +110,17 @@ class ChieriBotV4Rois:
first_rect = XYWHRect(x=self.left, y=self.top, w=self.width, h=self.height)
results = []
last_rect = first_rect
for vi in range(self.vertical_items):
rect = XYWHRect(*first_rect)
rect += (0, (self.vertical_gap + self.height) * vi, 0, 0)
for hi in range(self.horizontal_items):
if hi > 0:
rect += ((self.width + self.horizontal_gap), 0, 0, 0)
int_rect = construct_int_xywh_rect(rect)
results.append(crop_xywh(img_bgr, int_rect))
results.append(crop_xywh(img_bgr, rect.rounded()))
last_rect = rect
rect += (
last_rect += (
-(self.width + self.horizontal_gap) * 2,
self.height + self.b33_vertical_gap,
0,
@ -129,8 +128,7 @@ class ChieriBotV4Rois:
)
for hi in range(self.horizontal_items):
if hi > 0:
rect += ((self.width + self.horizontal_gap), 0, 0, 0)
int_rect = construct_int_xywh_rect(rect)
results.append(crop_xywh(img_bgr, int_rect))
last_rect += ((self.width + self.horizontal_gap), 0, 0, 0)
results.append(crop_xywh(img_bgr, last_rect.rounded()))
return results

View File

@ -1,25 +1,36 @@
from collections.abc import Iterable
from typing import NamedTuple, Tuple, Union
from math import floor
from typing import Callable, NamedTuple, Union
import numpy as np
Mat = np.ndarray
_IntOrFloat = Union[int, float]
class XYWHRect(NamedTuple):
x: int
y: int
w: int
h: int
x: _IntOrFloat
y: _IntOrFloat
w: _IntOrFloat
h: _IntOrFloat
def __add__(self, other: Union["XYWHRect", Tuple[int, int, int, int]]):
if not isinstance(other, Iterable) or len(other) != 4:
raise ValueError()
def _to_int(self, func: Callable[[_IntOrFloat], int]):
return (func(self.x), func(self.y), func(self.w), func(self.h))
def rounded(self):
return self._to_int(round)
def floored(self):
return self._to_int(floor)
def __add__(self, other):
if not isinstance(other, (list, tuple)) or len(other) != 4:
raise TypeError()
return self.__class__(*[a + b for a, b in zip(self, other)])
def __sub__(self, other: Union["XYWHRect", Tuple[int, int, int, int]]):
if not isinstance(other, Iterable) or len(other) != 4:
raise ValueError()
def __sub__(self, other):
if not isinstance(other, (list, tuple)) or len(other) != 4:
raise TypeError()
return self.__class__(*[a - b for a, b in zip(self, other)])

View File

@ -1,5 +1,5 @@
from collections.abc import Iterable
from typing import Callable, TypeVar, Union, overload
from typing import TypeVar, overload
import cv2
import numpy as np
@ -15,32 +15,25 @@ def imread_unicode(filepath: str, flags: int = cv2.IMREAD_UNCHANGED):
return cv2.imdecode(np.fromfile(filepath, dtype=np.uint8), flags)
def construct_int_xywh_rect(
rect: XYWHRect, func: Callable[[Union[int, float]], int] = round
):
return XYWHRect(*[func(num) for num in rect])
@overload
def apply_factor(item: int, factor: float) -> float: ...
@overload
def apply_factor(item: int, factor: float) -> float:
...
@overload
def apply_factor(item: float, factor: float) -> float:
...
def apply_factor(item: float, factor: float) -> float: ...
T = TypeVar("T", bound=Iterable)
@overload
def apply_factor(item: T, factor: float) -> T:
...
def apply_factor(item: T, factor: float) -> T: ...
def apply_factor(item, factor: float):
if isinstance(item, (int, float)):
return item * factor
if isinstance(item, XYWHRect):
return item.__class__(*[i * factor for i in item])
if isinstance(item, Iterable):
return item.__class__([i * factor for i in item])