PaddleOCR/v3.0.0/version3.x/module_usage/doc_img_orientation_classification

[giscus[bot]]

PaddleOCR/v3.0.0/version3.x/module_usage/doc_img_orientation_classification

Awesome multilingual OCR toolkits based on PaddlePaddle (practical ultra lightweight OCR system, support 80+ languages recognition, provide data annotation and synthesis tools, support training and deployment among server, mobile, embedded and IoT devices)

https://paddlepaddle.github.io/PaddleOCR/v3.0.0/version3.x/module_usage/doc_img_orientation_classification.html

[chandrakantpatel2403]

How to resolve below ERROR ??

2025-07-25 12:20:08,180 - INFO - Processing page 1/1 - Shape: (877, 621, 3)
2025-07-25 12:20:08,180 - INFO - Processing page 1/1 - Shape: (877, 621, 3)
2025-07-25 12:20:08,840 - ERROR - OCR failed on page 1 of Invoice_2526_0011.pdf: In user code:

InvalidArgumentError: The size of Op(Conv) inputs should not be 0.
  [Hint: Expected in_dims[i] != 0, but received in_dims[i]:0 == 0:0.] (at /paddle/paddle/phi/infermeta/binary.cc:589)
  [operator < pd_kernel.phi_kernel > error]

Traceback (most recent call last):
File "/app/ocr_engine.py", line 108, in process_images
result = ocr_engine(np_img)
File "/app/PaddleOCR/ppstructure/predict_system.py", line 130, in call
layout_res, elapse = self.layout_predictor(img)
File "/app/PaddleOCR/ppstructure/layout/predict_layout.py", line 96, in call
self.predictor.run()
ValueError: In user code:

InvalidArgumentError: The size of Op(Conv) inputs should not be 0.
  [Hint: Expected in_dims[i] != 0, but received in_dims[i]:0 == 0:0.] (at /paddle/paddle/phi/infermeta/binary.cc:589)
  [operator < pd_kernel.phi_kernel > error]

2025-07-25 12:20:08,840 - ERROR - OCR failed on page 1 of Invoice_2526_0011.pdf: In user code:

InvalidArgumentError: The size of Op(Conv) inputs should not be 0.
  [Hint: Expected in_dims[i] != 0, but received in_dims[i]:0 == 0:0.] (at /paddle/paddle/phi/infermeta/binary.cc:589)
  [operator < pd_kernel.phi_kernel > error]

Traceback (most recent call last):
File "/app/ocr_engine.py", line 108, in process_images
result = ocr_engine(np_img)
File "/app/PaddleOCR/ppstructure/predict_system.py", line 130, in call
layout_res, elapse = self.layout_predictor(img)
File "/app/PaddleOCR/ppstructure/layout/predict_layout.py", line 96, in call
self.predictor.run()
ValueError: In user code:

InvalidArgumentError: The size of Op(Conv) inputs should not be 0.
  [Hint: Expected in_dims[i] != 0, but received in_dims[i]:0 == 0:0.] (at /paddle/paddle/phi/infermeta/binary.cc:589)
  [operator < pd_kernel.phi_kernel > error]

2025-07-25 12:20:08,844 - ERROR - Unexpected error while processing Invoice_2526_0011.pdf: In user code:

InvalidArgumentError: The size of Op(Conv) inputs should not be 0.
  [Hint: Expected in_dims[i] != 0, but received in_dims[i]:0 == 0:0.] (at /paddle/paddle/phi/infermeta/binary.cc:589)
  [operator < pd_kernel.phi_kernel > error]

2025-07-25 12:20:08,844 - ERROR - Unexpected error while processing Invoice_2526_0011.pdf: In user code:

InvalidArgumentError: The size of Op(Conv) inputs should not be 0.
  [Hint: Expected in_dims[i] != 0, but received in_dims[i]:0 == 0:0.] (at /paddle/paddle/phi/infermeta/binary.cc:589)
  [operator < pd_kernel.phi_kernel > error]

2025-07-25 12:20:08,845 - ERROR - Traceback (most recent call last):
File "/app/app.py", line 127, in parse_invoice
all_results = process_images(images, ocr_engine, file.filename)
File "/app/ocr_engine.py", line 117, in process_images
raise e
File "/app/ocr_engine.py", line 108, in process_images
result = ocr_engine(np_img)
File "/app/PaddleOCR/ppstructure/predict_system.py", line 130, in call
layout_res, elapse = self.layout_predictor(img)
File "/app/PaddleOCR/ppstructure/layout/predict_layout.py", line 96, in call
self.predictor.run()
ValueError: In user code:

InvalidArgumentError: The size of Op(Conv) inputs should not be 0.
  [Hint: Expected in_dims[i] != 0, but received in_dims[i]:0 == 0:0.] (at /paddle/paddle/phi/infermeta/binary.cc:589)
  [operator < pd_kernel.phi_kernel > error]

2025-07-25 12:20:08,845 - ERROR - Traceback (most recent call last):
File "/app/app.py", line 127, in parse_invoice
all_results = process_images(images, ocr_engine, file.filename)
File "/app/ocr_engine.py", line 117, in process_images
raise e
File "/app/ocr_engine.py", line 108, in process_images
result = ocr_engine(np_img)
File "/app/PaddleOCR/ppstructure/predict_system.py", line 130, in call
layout_res, elapse = self.layout_predictor(img)
File "/app/PaddleOCR/ppstructure/layout/predict_layout.py", line 96, in call
self.predictor.run()
ValueError: In user code:

InvalidArgumentError: The size of Op(Conv) inputs should not be 0.
  [Hint: Expected in_dims[i] != 0, but received in_dims[i]:0 == 0:0.] (at /paddle/paddle/phi/infermeta/binary.cc:589)
  [operator < pd_kernel.phi_kernel > error]

2025-07-25 12:20:08,846 - INFO - 172.17.0.1:44076 - "POST /parse_invoice HTTP/1.1" 500

---

Configuration I have to used:

Device settings

args.use_gpu = False
args.use_npu = False
args.use_mlu = False
args.use_xpu = False
args.use_gcu = False
args.enable_mkldnn = False

# General
args.mode = "structure"
args.show_log = True
args.benchmark = False
args.output = "./output"
args.save_pdf = False
args.save_pdf_dir = None
args.image_dir = None
args.image_orientation = False
args.recovery = False

# Layout and formula detection (unless you need them)
args.layout = True
args.layout_model_dir = "/app/PaddleOCR/inference/layout/PP-DocLayout-L_infer" # picodet_lcnet_x1_0_fgd_layout_infer
args.layout_dict_path = "./PaddleOCR/ppocr/utils/dict/layout_dict/layout_doclayout_l_dict.txt" # layout_cdla_dict
args.layout_score_threshold = 0.5
args.layout_nms_threshold = 0.5
args.formula = False

# Table-specific settings
args.table = True
args.table_algorithm = "SLANet_plus"
args.table_model_dir = "/app/models/table/SLANet_plus_infer" # SLANet_infer
args.table_max_len = 2048
args.merge_no_span_structure = False
args.table_char_dict_path = "./PaddleOCR/ppocr/utils/dict/table_structure_dict_ch.txt"

# OCR backend
args.ocr = True
args.det_algorithm = "DB"
args.det_model_dir = "./PaddleOCR/inference/det/PP-OCRv5_server_det_infer"
args.det_limit_side_len = 960
args.det_limit_type = "max"
args.det_db_thresh = 0.3
args.det_db_box_thresh = 0.5
args.det_db_unclip_ratio = 1.5
args.use_dilation = False
args.det_db_score_mode = "fast"
args.det_box_type = "quad"
args.use_onnx = False

args.rec_model_dir = "/app/models/rec/en_PP-OCRv4_rec_infer"
args.rec_char_dict_path = "/app/PaddleOCR/ppocr/utils/en_dict.txt"
args.rec_algorithm = "CRNN"
args.rec_image_shape = "3,48,320"
args.rec_batch_num = 6
args.use_space_char = True
args.return_word_box = False
args.drop_score = 0.5

# Optional visual/debug settings
args.save_crop_res = False
args.return_ocr_result_in_table = True

# Angle classifier - optional
args.use_angle_cls = False
args.cls_model_dir = None # "/app/models/cls/ch_ppocr_mobile_v2.0_cls_infer"
args.cls_image_shape = "3,48,192"
args.cls_batch_num = 6
args.cls_thresh = 0.9
args.label_list = ['0', '180']

[Bobholamovic]

Please provide the paddlepaddle and paddleocr versions you are using.