飞桨黑客松 OpenVINO™ 任务获奖者经验分享 | 基于OpenVINO™ 与PaddleOCR的结构化输出Pipeline

# 配置坐标信息# The coordinates of the corners of the screen in case 1
POINTS = [[1121, 56],    # Left top          [3242, 183],   # right top          [3040, 1841],  # right bottom          [1000, 1543]]   # left bottom          # The size of the screen in case 1DESIGN_SHAPE = (1300, 1000)
# 配置坐标信息# The coordinates of the corners of the screen in case 1POINTS = [[1121, 56],    # Left top          [3242, 183],   # right top          [3040, 1841],  # right bottom          [1000, 1543]]   # left bottom
# The size of the screen in case 1DESIGN_SHAPE = (1300, 1000)
# 配置坐标信息# The coordinates of the corners of the screen in case 1POINTS = [[1121, 56],    # Left top          [3242, 183],   # right top          [3040, 1841],  # right bottom          [1000, 1543]]   # left bottom          # The size of the screen in case 1DESIGN_SHAPE = (1300, 1000)

from openvino.runtime import Core
# Initialize OpenVINO Runtime for text recognition.core = Core()
# Read the model and corresponding weights from a file.rec_model_file_path = "ch_PP-OCRv3_rec_infer/inference.pdmodel"rec_model = core.read_model(model=rec_model_file_path)
# Assign dynamic shapes to every input layer on the last dimension.for input_layer in rec_model.inputs:    input_shape = input_layer.partial_shape    input_shape[3] = -1    rec_model.reshape({input_layer: input_shape})
rec_compiled_model = core.compile_model(model=rec_model, device_name="CPU")
# Get input and output nodes.rec_input_layer = rec_compiled_model.input(0)rec_output_layer = rec_compiled_model.output(0)

# features and layout informationDESIGN_LAYOUT = {'Info_Probe':[14, 36, 410, 135],  # feature_name, xmin, ymin, xmax, ymax                 'Freq_Set':[5, 290, 544, 406],                 'Val_Total':[52, 419, 1256, 741],                 'Val_X':[19, 774, 433, 882],                 'Val_Y':[433, 773, 874, 884],                 'Val_Z':[873, 773, 1276, 883],                 'Unit':[1064, 291, 1295, 403],                 'Field':[5, 913, 243, 998]}

def prep_for_rec(dt_boxes, frame):    """    Preprocessing of the detected bounding boxes for text recognition
    Parameters:        dt_boxes: detected bounding boxes from text detection         frame: original input frame     """    ori_im = frame.copy()    img_crop_list = []     for bno in range(len(dt_boxes)):        tmp_box = copy.deepcopy(dt_boxes[bno])        img_crop = get_rotate_crop_image(ori_im, tmp_box)        img_crop_list.append(img_crop)            img_num = len(img_crop_list)    # 计算对应的缩放比例    width_list = []    for img in img_crop_list:        width_list.append(img.shape[1] / float(img.shape[0]))        # 对文字进行排序以加速推理    indices = np.argsort(np.array(width_list))    return img_crop_list, img_num, indices

def batch_text_box(img_crop_list, img_num, indices, beg_img_no, batch_num):    """    Batch for text recognition
    Parameters:        img_crop_list: processed bounding box images with detected bounding box        img_num: number of bounding boxes from text detection        indices: sorting for bounding boxes to speed up text recognition        beg_img_no: the beginning number of bounding boxes for each batch of text recognition inference        batch_num: number of images in each batch    """    norm_img_batch = []    max_wh_ratio = 0    end_img_no = min(img_num, beg_img_no + batch_num)    for ino in range(beg_img_no, end_img_no):        h, w = img_crop_list[indices[ino]].shape[0:2]        wh_ratio = w * 1.0 / h        max_wh_ratio = max(max_wh_ratio, wh_ratio)    for ino in range(beg_img_no, end_img_no):        norm_img = resize_norm_img(img_crop_list[indices[ino]], max_wh_ratio)        norm_img = norm_img[np.newaxis, :]        norm_img_batch.append(norm_img)
    norm_img_batch = np.concatenate(norm_img_batch)    norm_img_batch = norm_img_batch.copy()return norm_img_batch

class RecLabelDecode(object):    """ Convert between text-label and text-index """
    def __init__(self,                 character_dict_path=None,                 character_type='ch',                 use_space_char=False):        # 初始化函数
            def __call__(self, preds, label=None, *args, **kwargs):        # 对字符进行解码
# 解码函数，用于将id转换为对应地文字def decode(self, text_index, text_prob=None, is_remove_duplicate=False):        """ convert text-index into text-label. """        result_list = []        ignored_tokens = self.get_ignored_tokens()        batch_size = len(text_index)        for batch_idx in range(batch_size):            char_list = []            conf_list = []            for idx in range(len(text_index[batch_idx])):                # 将id转换为对应地文字            text = ''.join(char_list)            result_list.append((text, np.mean(conf_list)))        return result_list

# 输出结构体struct_result = {} 
# Crop imgs according the layout informationxmin, ymin, xmax, ymax = DESIGN_LAYOUT['Freq_Set']crop_img = img[ymin:ymax, xmin:xmax]
h = ymax - ymin  # height of crop_imgw = xmax - xmin  # width of crop_imgdt_boxes = [np.array([[0,0],[w,0],[w,h],[0,h]],dtype='float32')]batch_num = 1
# since the input img is cropped, we do not need a detection model to find the position of texts# Preprocess detection results for recognition.img_crop_list, img_num, indices = prep_for_rec(dt_boxes, crop_img)
# txts are the recognized text resultsrec_res = [['', 0.0]] * img_numtxts = [] 
for beg_img_no in range(0, img_num):
    # Recognition starts from here.    norm_img_batch = batch_text_box(        img_crop_list, img_num, indices, beg_img_no, batch_num)
    # Run inference for text recognition.     rec_results = rec_compiled_model([norm_img_batch])[rec_output_layer]
    # Postprocessing recognition results.    rec_result = text_decoder(rec_results)    for rno in range(len(rec_result)):        rec_res[indices[beg_img_no + rno]] = rec_result[rno]       if rec_res:        txts = [rec_res[i][0] for i in range(len(rec_res))] 
# record the recognition resultstruct_result['Freq_Set'] = txts[0]print(txts[0])

# Post-processing, fix some error made in recognitiondef post_processing(results, post_configration):    """    Postprocessing function for correcting the recognition errors.    Parameters:        results (Dict): The result directory.        post_configration (Dict): The configuration directory.    """    for key in results.keys():        if len(post_configration[key]) == 0:            continue  # nothing to do        for post_item in post_configration[key]:            key_word = post_item[0]            if key_word == 'MP':  # mapping                source_word = post_item[1]                target_word = post_item[2]                if source_word in results[key]:                    results[key] = target_word            elif key_word == 'RP':  # removing                source_word = post_item[1]                target_word = post_item[2]                results[key] = results[key].replace(source_word, target_word)            elif key_word == 'AD':  # add point                add_position = post_item[1]                results[key] = results[key][:add_position] + '.' + results[key][add_position:]    return results

# Congiguration for postprocessing of the resultsRESULT_POST = {"Info_Probe":[['MP', 'LF', '探头:LF-01']],  # words need to be mapped               "Freq_Set":[['RP', '实时值', ''], ['RP', ' ', ''], ['RP', 'H2', 'HZ']],  # words need to be replace               "Val_Total":[['RP', 'H2', 'Hz']],               "Val_X":[['RP', 'X', ''], ['RP', ':', '']],                "Val_Y":[['RP', 'Y', ''], ['RP', ':', '']],                "Val_Z":[['RP', 'Z', ''], ['RP', ':', '']],                "Unit":[['MP', 'T', 'μT'],['MP', 'kV', 'kV/m'],['MP', 'kv', 'kV/m'],['MP', 'vm', 'V/m'],['MP', 'Vm', 'V/m'],['MP', 'A', 'A/m']],                "Field":[]}  # nothing need to do
# Postprocessing, to fix some error made in recognitionstruct_result = post_processing(struct_result, RESULT_POST)
# Print resultprint(struct_result)

# 为了避免因为图片模糊导致的漏检，配置一个输出模板，从而让每个图片输出格式都一致# Output template in case 1RESULT_TEMP = {"Info_Probe":"探头:---",                "Freq_Set":"",                "Val_Total":"无探头",                "Val_X":"",                "Val_Y":"",                "Val_Z":"",                "Unit":"A/m",                "Field":"常规"}# the input of recognition should be image, DESIGN information, compiled_modeldef main_function(img, DESIGN_LAYOUT, RESULT_TEMP, preprocess_function=None):    # 根据配置信息获取输出结构的格式    struct_result = copy.deepcopy(RESULT_TEMP)
    # 逐个访问各个检测目标    for key in DESIGN_LAYOUT.keys():        # 裁剪各个区域的图片，并进行预处理# 对预处理后的图片进行识别        # 返回识别结果    return struct_result

# 读取图片并进行图片预处理img = cv2.imread('example1.jpg')img = pre_processing(img, POINTS, DESIGN_SHAPE)
# 进行文字识别struct_result = main_function(img, DESIGN_LAYOUT, RESULT_TEMP)
# 对识别结果进行精炼instruct_result = post_processing(struct_result, RESULT_POST)