2025.3.7 PaddleOCR 2.10 版本，预训练模型推理，报错 #14840

xiaozhangxiaozhang916 · 2025-03-12T03:08:47Z

xiaozhangxiaozhang916
Mar 12, 2025

[2025/03/12 10:56:35] ppocr WARNING: Skipping import of the encryption module.
[2025/03/12 10:56:35] ppocr INFO: Architecture :
[2025/03/12 10:56:35] ppocr INFO: Backbone :
[2025/03/12 10:56:35] ppocr INFO: model_name : small
[2025/03/12 10:56:35] ppocr INFO: name : MobileNetV3
[2025/03/12 10:56:35] ppocr INFO: scale : 0.35
[2025/03/12 10:56:35] ppocr INFO: Head :
[2025/03/12 10:56:35] ppocr INFO: class_dim : 2
[2025/03/12 10:56:35] ppocr INFO: name : ClsHead
[2025/03/12 10:56:35] ppocr INFO: Neck : None
[2025/03/12 10:56:35] ppocr INFO: Transform : None
[2025/03/12 10:56:35] ppocr INFO: algorithm : CLS
[2025/03/12 10:56:35] ppocr INFO: model_type : cls
[2025/03/12 10:56:35] ppocr INFO: Eval :
[2025/03/12 10:56:35] ppocr INFO: dataset :
[2025/03/12 10:56:35] ppocr INFO: data_dir : ./train_data/cls
[2025/03/12 10:56:35] ppocr INFO: label_file_list : ['./train_data/cls/test.txt']
[2025/03/12 10:56:35] ppocr INFO: name : SimpleDataSet
[2025/03/12 10:56:35] ppocr INFO: transforms :
[2025/03/12 10:56:35] ppocr INFO: DecodeImage :
[2025/03/12 10:56:35] ppocr INFO: channel_first : False
[2025/03/12 10:56:35] ppocr INFO: img_mode : BGR
[2025/03/12 10:56:35] ppocr INFO: ClsLabelEncode : None
[2025/03/12 10:56:35] ppocr INFO: ClsResizeImg :
[2025/03/12 10:56:35] ppocr INFO: image_shape : [3, 48, 192]
[2025/03/12 10:56:35] ppocr INFO: KeepKeys :
[2025/03/12 10:56:35] ppocr INFO: keep_keys : ['image', 'label']
[2025/03/12 10:56:35] ppocr INFO: loader :
[2025/03/12 10:56:35] ppocr INFO: batch_size_per_card : 512
[2025/03/12 10:56:35] ppocr INFO: drop_last : False
[2025/03/12 10:56:35] ppocr INFO: num_workers : 4
[2025/03/12 10:56:35] ppocr INFO: shuffle : False
[2025/03/12 10:56:35] ppocr INFO: Global :
[2025/03/12 10:56:35] ppocr INFO: cal_metric_during_train : True
[2025/03/12 10:56:35] ppocr INFO: checkpoints : None
[2025/03/12 10:56:35] ppocr INFO: distributed : False
[2025/03/12 10:56:35] ppocr INFO: epoch_num : 100
[2025/03/12 10:56:35] ppocr INFO: eval_batch_step : [0, 1000]
[2025/03/12 10:56:35] ppocr INFO: infer_img : doc/imgs_words/ch/word_1.jpg
[2025/03/12 10:56:35] ppocr INFO: label_list : ['0', '180']
[2025/03/12 10:56:35] ppocr INFO: load_static_weights : False
[2025/03/12 10:56:35] ppocr INFO: log_smooth_window : 20
[2025/03/12 10:56:35] ppocr INFO: pretrained_model : D:\zhangqian\DL\PaddleOCR-main/inference/PP-LCNet_x0_25_textline_ori_pretrained.pdparams
[2025/03/12 10:56:35] ppocr INFO: print_batch_step : 10
[2025/03/12 10:56:35] ppocr INFO: save_epoch_step : 3
[2025/03/12 10:56:35] ppocr INFO: save_inference_dir : None
[2025/03/12 10:56:35] ppocr INFO: save_model_dir : ./output/cls/mv3/
[2025/03/12 10:56:35] ppocr INFO: use_gpu : True
[2025/03/12 10:56:35] ppocr INFO: use_visualdl : False
[2025/03/12 10:56:35] ppocr INFO: Loss :
[2025/03/12 10:56:35] ppocr INFO: name : ClsLoss
[2025/03/12 10:56:35] ppocr INFO: Metric :
[2025/03/12 10:56:35] ppocr INFO: main_indicator : acc
[2025/03/12 10:56:35] ppocr INFO: name : ClsMetric
[2025/03/12 10:56:35] ppocr INFO: Optimizer :
[2025/03/12 10:56:35] ppocr INFO: beta1 : 0.9
[2025/03/12 10:56:35] ppocr INFO: beta2 : 0.999
[2025/03/12 10:56:35] ppocr INFO: lr :
[2025/03/12 10:56:35] ppocr INFO: learning_rate : 0.001
[2025/03/12 10:56:35] ppocr INFO: name : Cosine
[2025/03/12 10:56:35] ppocr INFO: name : Adam
[2025/03/12 10:56:35] ppocr INFO: regularizer :
[2025/03/12 10:56:35] ppocr INFO: factor : 0
[2025/03/12 10:56:35] ppocr INFO: name : L2
[2025/03/12 10:56:35] ppocr INFO: PostProcess :
[2025/03/12 10:56:35] ppocr INFO: name : ClsPostProcess
[2025/03/12 10:56:35] ppocr INFO: Train :
[2025/03/12 10:56:35] ppocr INFO: dataset :
[2025/03/12 10:56:35] ppocr INFO: data_dir : ./train_data/cls
[2025/03/12 10:56:35] ppocr INFO: label_file_list : ['./train_data/cls/train.txt']
[2025/03/12 10:56:35] ppocr INFO: name : SimpleDataSet
[2025/03/12 10:56:35] ppocr INFO: transforms :
[2025/03/12 10:56:35] ppocr INFO: DecodeImage :
[2025/03/12 10:56:35] ppocr INFO: channel_first : False
[2025/03/12 10:56:35] ppocr INFO: img_mode : BGR
[2025/03/12 10:56:35] ppocr INFO: ClsLabelEncode : None
[2025/03/12 10:56:35] ppocr INFO: BaseDataAugmentation : None
[2025/03/12 10:56:35] ppocr INFO: RandAugment : None
[2025/03/12 10:56:35] ppocr INFO: ClsResizeImg :
[2025/03/12 10:56:35] ppocr INFO: image_shape : [3, 48, 192]
[2025/03/12 10:56:35] ppocr INFO: KeepKeys :
[2025/03/12 10:56:35] ppocr INFO: keep_keys : ['image', 'label']
[2025/03/12 10:56:35] ppocr INFO: loader :
[2025/03/12 10:56:35] ppocr INFO: batch_size_per_card : 512
[2025/03/12 10:56:35] ppocr INFO: drop_last : True
[2025/03/12 10:56:35] ppocr INFO: num_workers : 8
[2025/03/12 10:56:35] ppocr INFO: shuffle : True
[2025/03/12 10:56:35] ppocr INFO: profiler_options : None
[2025/03/12 10:56:35] ppocr INFO: train with paddle 2.5.0-rc1 and device Place(gpu:0)
W0312 10:56:35.161926 7472 gpu_resources.cc:119] Please NOTE: device: 0, GPU Compute Capability: 8.9, Driver API Version: 12.4, Runtime API Version: 11.8
W0312 10:56:35.163925 7472 gpu_resources.cc:149] device: 0, cuDNN Version: 8.4.
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params conv1.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params conv1.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params conv1.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params conv1.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params conv1.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks2.0.dw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks2.0.dw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks2.0.dw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks2.0.dw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks2.0.dw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks2.0.pw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks2.0.pw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks2.0.pw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks2.0.pw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks2.0.pw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.0.dw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.0.dw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.0.dw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.0.dw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.0.dw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.0.pw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.0.pw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.0.pw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.0.pw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.0.pw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.1.dw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.1.dw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.1.dw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.1.dw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.1.dw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.1.pw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.1.pw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.1.pw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.1.pw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks3.1.pw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.0.dw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.0.dw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.0.dw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.0.dw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.0.dw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.0.pw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.0.pw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.0.pw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.0.pw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.0.pw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.1.dw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.1.dw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.1.dw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.1.dw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.1.dw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.1.pw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.1.pw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.1.pw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.1.pw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks4.1.pw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.0.dw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.0.dw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.0.dw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.0.dw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.0.dw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.0.pw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.0.pw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.0.pw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.0.pw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.0.pw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.1.dw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.1.dw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.1.dw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.1.dw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.1.dw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.1.pw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.1.pw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.1.pw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.1.pw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.1.pw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.2.dw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.2.dw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.2.dw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.2.dw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.2.dw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.2.pw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.2.pw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.2.pw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.2.pw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.2.pw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.3.dw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.3.dw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.3.dw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.3.dw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.3.dw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.3.pw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.3.pw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.3.pw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.3.pw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.3.pw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.4.dw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.4.dw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.4.dw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.4.dw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.4.dw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.4.pw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.4.pw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.4.pw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.4.pw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.4.pw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.5.dw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.5.dw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.5.dw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.5.dw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.5.dw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.5.pw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.5.pw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.5.pw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.5.pw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks5.5.pw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.0.dw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.0.dw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.0.dw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.0.dw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.0.dw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.0.se.conv1.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.0.se.conv1.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.0.se.conv2.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.0.se.conv2.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.0.pw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.0.pw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.0.pw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.0.pw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.0.pw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.1.dw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.1.dw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.1.dw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.1.dw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.1.dw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.1.se.conv1.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.1.se.conv1.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.1.se.conv2.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.1.se.conv2.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.1.pw_conv.conv.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.1.pw_conv.bn.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.1.pw_conv.bn.bias not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.1.pw_conv.bn._mean not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params blocks6.1.pw_conv.bn._variance not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params fc.weight not in model
[2025/03/12 10:56:35] ppocr WARNING: The pretrained params fc.bias not in model

GreatV · 2025-03-12T03:09:42Z

GreatV
Mar 12, 2025
Maintainer

该错误表明您加载的预训练模型参数与当前模型结构不匹配，导致多个参数无法正确加载。以下是可能的原因和解决方案：

可能的原因

预训练模型与当前模型架构不兼容
- 您当前使用的预训练模型 PP-LCNet_x0_25_textline_ori_pretrained.pdparams 可能与 MobileNetV3 small 结构不匹配，因此出现大量 WARNING: The pretrained params XXX not in model。
- 需要确认您的 Backbone 结构是否与预训练模型匹配。
scale 参数不同
- 预训练模型可能是不同 scale（如 x0_5 或 x1_0），而您的 config 文件中 scale: 0.35，导致参数不匹配。
- 请确保 scale 和 model_name 与预训练模型一致。
命令行错误导致未正确加载参数
- 在 Global.pretrained_model 路径是否正确（需要 .pdparams 文件，而不是 inference 目录）。
- 确保 load_static_weights 参数是否正确，PaddleOCR 新版本通常使用动态图权重。

解决方案

1. 更换正确的预训练模型

先检查 PP-LCNet_x0_25_textline_ori_pretrained.pdparams 是否为 MobileNetV3 small 的正确模型，而不是 PP-LCNet 版本。例如：
```
wget https://paddleocr.bj.bcebos.com/pretrained/paddleocr/ch_ppocr_mobile_v2.0_cls_train.pdparams
```
然后尝试使用 ch_ppocr_mobile_v2.0_cls_train.pdparams 作为预训练模型。

2. 确保 `scale` 版本匹配

如果当前 scale 为 0.35，但预训练模型是 x0.25 或 x0.5，则应：
- 修改 config 文件，使 scale 与预训练模型匹配：
```
Backbone:
    name: MobileNetV3
    scale: 0.25 # 或0.5，确保与下载的模型匹配
```
- 或下载 scale=0.35 对应的 PP-LCNet 预训练模型。

3. 直接忽略不匹配的参数

如果确定可以忽略部分错误，可在 train.py 启动时加上：

python tools/train.py --config configs/xxx.yml -o Global.pretrained_model=your_model_path.pdparams -o Global.ignore_weights=True

但这仅适用于 Head 层结构有变更的情况，不适用于 Backbone 层变更。

4. 检查 `load_static_weights` 设置

如果您的 .pdparams 是动态图权重，确保 load_static_weights: False。

总结

检查 Backbone 结构是否匹配 预训练模型。
确保 scale 版本与预训练模型一致，如 x0_25、x0_5 等。
下载正确的 .pdparams 预训练模型 并确保路径正确。
尝试忽略不匹配的参数（Global.ignore_weights=True），但不建议。

如果问题依然存在，请提供完整的 config 文件，我们可以进一步分析。

Response generated by 🤖 feifei-bot | chatgpt-4o-latest

⚠️ This is an AI-generated response. Please verify any critical information.

0 replies

xiaozhangxiaozhang916 · 2025-03-12T03:13:47Z

xiaozhangxiaozhang916
Mar 12, 2025
Author

config文件paddle2.10里面只有cls_mv3.yml一个可用文件

0 replies

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

2025.3.7 PaddleOCR 2.10 版本，预训练模型推理，报错 #14840

Uh oh!

{{title}}

Uh oh!

Replies: 2 comments

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Select a reply

Uh oh!

2025.3.7 PaddleOCR 2.10 版本，预训练模型推理，报错 #14840

Uh oh!

xiaozhangxiaozhang916 Mar 12, 2025

Replies: 2 comments

Uh oh!

GreatV Mar 12, 2025 Maintainer

可能的原因

解决方案

1. 更换正确的预训练模型

2. 确保 scale 版本匹配

3. 直接忽略不匹配的参数

4. 检查 load_static_weights 设置

相关讨论

总结

Uh oh!

xiaozhangxiaozhang916 Mar 12, 2025 Author

xiaozhangxiaozhang916
Mar 12, 2025

GreatV
Mar 12, 2025
Maintainer

2. 确保 `scale` 版本匹配

4. 检查 `load_static_weights` 设置

xiaozhangxiaozhang916
Mar 12, 2025
Author