π Awesome OCR multiple programing languages toolkits based on ONNX Runtime, OpenVINO, MNN, PaddlePaddle, TensorRT and PyTorch.
Optim ort infer
SWHLcdf5b031c8cb
diff --git a/python/rapidocr_onnxruntime/ch_ppocr_v2_cls/text_cls.py b/python/rapidocr_onnxruntime/ch_ppocr_v2_cls/text_cls.pyindex 3d2b673..a46b77e 100644--- a/python/rapidocr_onnxruntime/ch_ppocr_v2_cls/text_cls.py+++ b/python/rapidocr_onnxruntime/ch_ppocr_v2_cls/text_cls.py@@ -33,9 +33,7 @@ class TextClassifier():self.cls_thresh = config['cls_thresh']self.postprocess_op = ClsPostProcess(config['label_list'])- session_instance = OrtInferSession(config)- self.session = session_instance.session- self.input_name = session_instance.get_input_name()+ self.infer = OrtInferSession(config)def __call__(self, img_list: List[np.ndarray]):if isinstance(img_list, np.ndarray):@@ -64,8 +62,7 @@ class TextClassifier():norm_img_batch = np.concatenate(norm_img_batch).astype(np.float32)starttime = time.time()- onnx_inputs = {self.input_name: norm_img_batch}- prob_out = self.session.run(None, onnx_inputs)[0]+ prob_out = self.infer(norm_img_batch)[0]cls_result = self.postprocess_op(prob_out)elapse += time.time() - starttimediff --git a/python/rapidocr_onnxruntime/ch_ppocr_v2_cls/utils.py b/python/rapidocr_onnxruntime/ch_ppocr_v2_cls/utils.pyindex d37221e..d2df500 100644--- a/python/rapidocr_onnxruntime/ch_ppocr_v2_cls/utils.py+++ b/python/rapidocr_onnxruntime/ch_ppocr_v2_cls/utils.py@@ -15,6 +15,7 @@ from pathlib import Pathimport warningsimport yaml+import numpy as npfrom onnxruntime import (get_available_providers, get_device,SessionOptions, InferenceSession,GraphOptimizationLevel)@@ -53,11 +54,18 @@ class OrtInferSession():'https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html',RuntimeWarning)- def get_input_name(self, input_idx=0):- return self.session.get_inputs()[input_idx].name+ def __call__(self, input_content: np.ndarray) -> np.ndarray:+ input_dict = dict(zip(self.get_input_names(), input_content))+ try:+ return self.session.run(self.get_output_names(), input_dict)+ except Exception as e:+ raise ONNXRuntimeError('ONNXRuntime inferece failed.') from e- def get_output_name(self, output_idx=0):- return self.session.get_outputs()[output_idx].name+ def get_input_names(self, ):+ return [v.name for v in self.session.get_inputs()]++ def get_output_names(self,):+ return [v.name for v in self.session.get_outputs()]@staticmethoddef _verify_model(model_path):@@ -68,6 +76,10 @@ class OrtInferSession():raise FileExistsError(f'{model_path} is not a file.')+class ONNXRuntimeError(Exception):+ pass++def read_yaml(yaml_path):with open(yaml_path, 'rb') as f:data = yaml.load(f, Loader=yaml.Loader)diff --git a/python/rapidocr_onnxruntime/ch_ppocr_v3_det/text_detect.py b/python/rapidocr_onnxruntime/ch_ppocr_v3_det/text_detect.pyindex c3ebf6e..38cd186 100644--- a/python/rapidocr_onnxruntime/ch_ppocr_v3_det/text_detect.py+++ b/python/rapidocr_onnxruntime/ch_ppocr_v3_det/text_detect.py@@ -33,11 +33,9 @@ class TextDetector():self.preprocess_op = create_operators(config['pre_process'])self.postprocess_op = DBPostProcess(**config['post_process'])- session_instance = OrtInferSession(config)- self.session = session_instance.session- self.input_name = session_instance.get_input_name()+ self.infer = OrtInferSession(config)- def __call__(self, img):+ def __call__(self, img: np.ndarry):if img is None:raise ValueError('img is None')@@ -53,9 +51,8 @@ class TextDetector():shape_list = np.expand_dims(shape_list, axis=0)starttime = time.time()- preds = self.session.run(None, {self.input_name: img})-- post_result = self.postprocess_op(preds[0], shape_list)+ preds = self.infer(img)[0]+ post_result = self.postprocess_op(preds, shape_list)dt_boxes = post_result[0]['points']dt_boxes = self.filter_tag_det_res(dt_boxes, ori_im_shape)diff --git a/python/rapidocr_onnxruntime/ch_ppocr_v3_det/utils.py b/python/rapidocr_onnxruntime/ch_ppocr_v3_det/utils.pyindex dc7eccd..110c14d 100644--- a/python/rapidocr_onnxruntime/ch_ppocr_v3_det/utils.py+++ b/python/rapidocr_onnxruntime/ch_ppocr_v3_det/utils.py@@ -64,11 +64,18 @@ class OrtInferSession():'https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html',RuntimeWarning)- def get_input_name(self, input_idx=0):- return self.session.get_inputs()[input_idx].name+ def __call__(self, input_content: np.ndarray) -> np.ndarray:+ input_dict = dict(zip(self.get_input_names(), input_content))+ try:+ return self.session.run(self.get_output_names(), input_dict)+ except Exception as e:+ raise ONNXRuntimeError('ONNXRuntime inferece failed.') from e++ def get_input_names(self, )+ return [v.name for v in self.session.get_inputs()]- def get_output_name(self, output_idx=0):- return self.session.get_outputs()[output_idx].name+ def get_output_names(self,):+ return [v.name for v in self.session.get_outputs()]@staticmethoddef _verify_model(model_path):@@ -79,6 +86,10 @@ class OrtInferSession():raise FileExistsError(f'{model_path} is not a file.')+class ONNXRuntimeError(Exception):+ pass++def read_yaml(yaml_path):with open(yaml_path, 'rb') as f:data = yaml.load(f, Loader=yaml.Loader)diff --git a/python/rapidocr_onnxruntime/ch_ppocr_v3_rec/text_recognize.py b/python/rapidocr_onnxruntime/ch_ppocr_v3_rec/text_recognize.pyindex 116eabb..e996420 100644--- a/python/rapidocr_onnxruntime/ch_ppocr_v3_rec/text_recognize.py+++ b/python/rapidocr_onnxruntime/ch_ppocr_v3_rec/text_recognize.py@@ -27,12 +27,10 @@ except:class TextRecognizer():def __init__(self, config):- session_instance = OrtInferSession(config)- self.session = session_instance.session- self.input_name = session_instance.get_input_name()+ self.session = OrtInferSession(config)- if session_instance.have_key():- self.character_dict_path = session_instance.get_character_list()+ if self.session.have_key():+ self.character_dict_path = self.session.get_character_list()else:self.character_dict_path = config.get('keys_path', None)self.postprocess_op = CTCLabelDecode(self.character_dict_path)@@ -71,8 +69,7 @@ class TextRecognizer():norm_img_batch = np.concatenate(norm_img_batch).astype(np.float32)starttime = time.time()- onnx_inputs = {self.input_name: norm_img_batch}- preds = self.session.run(None, onnx_inputs)[0]+ preds = self.session(norm_img_batch)[0]rec_result = self.postprocess_op(preds)for rno in range(len(rec_result)):diff --git a/python/rapidocr_onnxruntime/ch_ppocr_v3_rec/utils.py b/python/rapidocr_onnxruntime/ch_ppocr_v3_rec/utils.pyindex 8d04f25..9198df4 100644--- a/python/rapidocr_onnxruntime/ch_ppocr_v3_rec/utils.py+++ b/python/rapidocr_onnxruntime/ch_ppocr_v3_rec/utils.py@@ -44,11 +44,18 @@ class OrtInferSession():'https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html',RuntimeWarning)- def get_input_name(self, input_idx=0):- return self.session.get_inputs()[input_idx].name+ def __call__(self, input_content: np.ndarray) -> np.ndarray:+ input_dict = dict(zip(self.get_input_names(), input_content))+ try:+ return self.session.run(self.get_output_names(), input_dict)+ except Exception as e:+ raise ONNXRuntimeError('ONNXRuntime inferece failed.') from e- def get_output_name(self, output_idx=0):- return self.session.get_outputs()[output_idx].name+ def get_input_names(self, ):+ return [v.name for v in self.session.get_inputs()]++ def get_output_names(self,):+ return [v.name for v in self.session.get_outputs()]def get_character_list(self, key: str = 'character'):return self.meta_dict[key].splitlines()@@ -68,6 +75,10 @@ class OrtInferSession():raise FileExistsError(f'{model_path} is not a file.')+class ONNXRuntimeError(Exception):+ pass++def read_yaml(yaml_path):with open(yaml_path, 'rb') as f:data = yaml.load(f, Loader=yaml.Loader)diff --git a/python/tests/test_all_ort.py b/python/tests/test_all_ort.pyindex 1a77900..be68e7a 100644--- a/python/tests/test_all_ort.py+++ b/python/tests/test_all_ort.py@@ -15,7 +15,7 @@ sys.path.append(str(root_dir))from rapidocr_onnxruntime import RapidOCRrapid_ocr = RapidOCR()-tests_dir = root_dir / 'test_files'+tests_dir = root_dir / 'tests' / 'test_files'def test_normal():diff --git a/python/tests/test_all_vino.py b/python/tests/test_all_vino.pyindex fe1cca0..479fea0 100644--- a/python/tests/test_all_vino.py+++ b/python/tests/test_all_vino.py@@ -16,7 +16,7 @@ from rapidocr_openvino import RapidOCRrapid_ocr = RapidOCR()-tests_dir = root_dir / 'test_files'+tests_dir = root_dir / 'tests' / 'test_files'def test_normal():