opencv/text_recognition_crnn

1# This file is part of OpenCV Zoo project.2# It is subject to the license terms in the LICENSE file found in the same directory.3#4# Copyright (C) 2021, Shenzhen Institute of Artificial Intelligence and Robotics for Society, all rights reserved.5# Third party copyrights are property of their respective owners.6 7import numpy as np8import cv2 as cv9 10class PPOCRDet:11    def __init__(self, modelPath, inputSize=[736, 736], binaryThreshold=0.3, polygonThreshold=0.5, maxCandidates=200, unclipRatio=2.0, backendId=0, targetId=0):12        self._modelPath = modelPath13        self._model = cv.dnn_TextDetectionModel_DB(14            cv.dnn.readNet(self._modelPath)15        )16 17        self._inputSize = tuple(inputSize) # (w, h)18        self._inputHeight = inputSize[0]19        self._inputWidth = inputSize[1]20        self._binaryThreshold = binaryThreshold21        self._polygonThreshold = polygonThreshold22        self._maxCandidates = maxCandidates23        self._unclipRatio = unclipRatio24        self._backendId = backendId25        self._targetId = targetId26 27        self._model.setPreferableBackend(self._backendId)28        self._model.setPreferableTarget(self._targetId)29 30        self._model.setBinaryThreshold(self._binaryThreshold)31        self._model.setPolygonThreshold(self._polygonThreshold)32        self._model.setUnclipRatio(self._unclipRatio)33        self._model.setMaxCandidates(self._maxCandidates)34 35        self._model.setInputSize(self._inputSize)36        self._model.setInputMean((123.675, 116.28, 103.53))37        self._model.setInputScale(1.0/255.0/np.array([0.229, 0.224, 0.225]))38 39    @property40    def name(self):41        return self.__class__.__name__42 43    def setBackendAndTarget(self, backendId, targetId):44        self._backendId = backendId45        self._targetId = targetId46        self._model.setPreferableBackend(self._backendId)47        self._model.setPreferableTarget(self._targetId)48 49    def setInputSize(self, input_size):50        self._inputSize = tuple(input_size)51        self._model.setInputSize(self._inputSize)52        self._model.setInputMean((123.675, 116.28, 103.53))53        self._model.setInputScale(1.0/255.0/np.array([0.229, 0.224, 0.225]))54 55    def infer(self, image):56        assert image.shape[0] == self._inputSize[1], '{} (height of input image) != {} (preset height)'.format(image.shape[0], self._inputSize[1])57        assert image.shape[1] == self._inputSize[0], '{} (width of input image) != {} (preset width)'.format(image.shape[1], self._inputSize[0])58 59        return self._model.detect(image)60