π Awesome OCR multiple programing languages toolkits based on ONNX Runtime, OpenVINO, MNN, PaddlePaddle, TensorRT and PyTorch.
# -*- encoding: utf-8 -*-
# @Author: SWHL
# @Contact: [email protected]
from typing import Any, Dict, List, Tuple
import cv2
import numpy as np
def map_img_to_original(
imgs: List[np.ndarray], ratio_h: float, ratio_w: float
) -> List[np.ndarray]:
results = []
for img in imgs:
img_h, img_w = img.shape[:2]
ori_img_h, ori_img_w = round(img_h * ratio_h), round(img_w * ratio_w)
resize_img = cv2.resize(img, (ori_img_w, ori_img_h))
results.append(resize_img)
return results
def map_boxes_to_original(
dt_boxes: np.ndarray, op_record: Dict[str, Any], ori_h: int, ori_w: int
) -> np.ndarray:
for op in reversed(list(op_record.keys())):
v = op_record[op]
if "padding" in op:
top, left = v.get("top"), v.get("left")
dt_boxes[:, :, 0] -= left
dt_boxes[:, :, 1] -= top
elif "preprocess" in op:
ratio_h = v.get("ratio_h")
ratio_w = v.get("ratio_w")
dt_boxes[:, :, 0] *= ratio_w
dt_boxes[:, :, 1] *= ratio_h
dt_boxes = np.where(dt_boxes < 0, 0, dt_boxes)
dt_boxes[..., 0] = np.where(dt_boxes[..., 0] > ori_w, ori_w, dt_boxes[..., 0])
dt_boxes[..., 1] = np.where(dt_boxes[..., 1] > ori_h, ori_h, dt_boxes[..., 1])
return dt_boxes
def apply_vertical_padding(
img: np.ndarray,
op_record: Dict[str, Any],
width_height_ratio: float,
min_height: float,
) -> Tuple[np.ndarray, Dict[str, Any]]:
h, w = img.shape[:2]
if width_height_ratio == -1:
use_limit_ratio = False
else:
use_limit_ratio = w / h > width_height_ratio
if h <= min_height or use_limit_ratio:
padding_h = get_padding_h(h, w, width_height_ratio, min_height)
block_img = add_round_letterbox(img, (padding_h, padding_h, 0, 0))
op_record["padding_1"] = {"top": padding_h, "left": 0}
return block_img, op_record
op_record["padding_1"] = {"top": 0, "left": 0}
return img, op_record
def get_padding_h(h: int, w: int, width_height_ratio: float, min_height: float) -> int:
new_h = max(int(w / width_height_ratio), min_height) * 2
padding_h = int(abs(new_h - h) / 2)
return padding_h
def get_rotate_crop_image(img: np.ndarray, points: np.ndarray) -> np.ndarray:
img_crop_width = int(
max(
np.linalg.norm(points[0] - points[1]),
np.linalg.norm(points[2] - points[3]),
)
)
img_crop_height = int(
max(
np.linalg.norm(points[0] - points[3]),
np.linalg.norm(points[1] - points[2]),
)
)
pts_std = np.array(
[
[0, 0],
[img_crop_width, 0],
[img_crop_width, img_crop_height],
[0, img_crop_height],
]
).astype(np.float32)
M = cv2.getPerspectiveTransform(points, pts_std)
dst_img = cv2.warpPerspective(
img,
M,
(img_crop_width, img_crop_height),
borderMode=cv2.BORDER_REPLICATE,
flags=cv2.INTER_CUBIC,
)
dst_img_height, dst_img_width = dst_img.shape[0:2]
if dst_img_height * 1.0 / dst_img_width >= 1.5:
dst_img = np.rot90(dst_img)
return dst_img
def resize_image_within_bounds(
img: np.ndarray, min_side_len: float, max_side_len: float
) -> Tuple[np.ndarray, float, float]:
original_h, original_w = img.shape[:2]
h, w = original_h, original_w
max_value = max(h, w)
if max_value > max_side_len:
img, _, _ = reduce_max_side(img, max_side_len)
h, w = img.shape[:2]
min_value = min(h, w)
if min_value < min_side_len:
img, _, _ = increase_min_side(img, min_side_len)
return img, original_h / img.shape[0], original_w / img.shape[1]
def reduce_max_side(
img: np.ndarray, max_side_len: float = 2000
) -> Tuple[np.ndarray, float, float]:
h, w = img.shape[:2]
ratio = 1.0
if max(h, w) > max_side_len:
if h > w:
ratio = float(max_side_len) / h
else:
ratio = float(max_side_len) / w
resize_h = int(h * ratio)
resize_w = int(w * ratio)
resize_h = int(round(resize_h / 32) * 32)
resize_w = int(round(resize_w / 32) * 32)
try:
if int(resize_w) <= 0 or int(resize_h) <= 0:
raise ResizeImgError("resize_w or resize_h is less than or equal to 0")
img = cv2.resize(img, (resize_w, resize_h))
except Exception as exc:
raise ResizeImgError() from exc
ratio_h = h / resize_h
ratio_w = w / resize_w
return img, ratio_h, ratio_w
def increase_min_side(
img: np.ndarray, min_side_len: float = 30
) -> Tuple[np.ndarray, float, float]:
h, w = img.shape[:2]
ratio = 1.0
if min(h, w) < min_side_len:
if h < w:
ratio = float(min_side_len) / h
else:
ratio = float(min_side_len) / w
resize_h = int(h * ratio)
resize_w = int(w * ratio)
resize_h = int(round(resize_h / 32) * 32)
resize_w = int(round(resize_w / 32) * 32)
try:
if int(resize_w) <= 0 or int(resize_h) <= 0:
raise ResizeImgError("resize_w or resize_h is less than or equal to 0")
img = cv2.resize(img, (resize_w, resize_h))
except Exception as exc:
raise ResizeImgError() from exc
ratio_h = h / resize_h
ratio_w = w / resize_w
return img, ratio_h, ratio_w
def add_round_letterbox(
img: np.ndarray, padding_tuple: Tuple[int, int, int, int]
) -> np.ndarray:
padded_img = cv2.copyMakeBorder(
img,
padding_tuple[0],
padding_tuple[1],
padding_tuple[2],
padding_tuple[3],
cv2.BORDER_CONSTANT,
value=(0, 0, 0),
)
return padded_img
class ResizeImgError(Exception):
pass