Created by: zhangmianhongni

最新V1.1的模型ch_ppocr_server_v1.1_xx开启了enable_mkldnn后会报错，报错信息：

File "D:/DIG/daas-dmg-ocr-py/tests/ocr/ocr_test.py", line 20, in ocr = PaddleOCR(det_model_dir=det_model_dir, rec_model_dir=rec_model_dir, cls_model_dir=cls_model_dir, use_angle_cls=True, lang='ch', use_gpu=False, enable_mkldnn=True) File "C:\Program Files\Python36\lib\site-packages\paddleocr\paddleocr.py", line 222, in init super().init(postprocess_params) File "C:\Program Files\Python36\lib\site-packages\paddleocr\tools\infer\predict_system.py", line 41, in init self.text_detector = predict_det.TextDetector(args) File "C:\Program Files\Python36\lib\site-packages\paddleocr\tools\infer\predict_det.py", line 80, in init utility.create_predictor(args, mode="det") File "C:\Program Files\Python36\lib\site-packages\paddleocr\tools\infer\utility.py", line 133, in create_predictor predictor = create_paddle_predictor(config) paddle.fluid.core_avx.EnforceNotMet:

C++ Call Stacks (More useful to developers):

Windows not support stack backtrace yet.

Error Message Summary:

Error: Attribute output_padding is not found in proto conv2d_transpose at (D:\1.8.4\paddle\paddle\fluid\framework\op_desc.cc:521)

如果使用GPU是没问题的，但是CPU+开启mkldnn就会报这个错，之前旧的版本使用ch_det_r50_vd_db和ch_rec_r34_vd_crnn模型的话是没问题的