search

PaddlePaddle / PaddleOCR

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/
Apache License 2.0
44.1k stars 7.81k forks source link

Process finished with exit code -1073741819 (0xC0000005) #2419

Closed DenisaLicu closed 3 years ago

DenisaLicu commented 3 years ago

I am trying to fine tune the model using the pre-trained model and a custom character dictionary. When loading the dataset I get this error, that I couldn't figure it out: Process finished with exit code -1073741819 (0xC0000005). I only have 6 images in the training data so a memory overload wouldn't be possible.

This is the complete output:

Architecture : [2021/04/08 16:41:30] root INFO: Backbone : [2021/04/08 16:41:30] root INFO: model_name : large [2021/04/08 16:41:30] root INFO: name : MobileNetV3 [2021/04/08 16:41:30] root INFO: scale : 0.5 [2021/04/08 16:41:30] root INFO: Head : [2021/04/08 16:41:30] root INFO: fc_decay : 0 [2021/04/08 16:41:30] root INFO: name : CTCHead [2021/04/08 16:41:30] root INFO: Neck : [2021/04/08 16:41:30] root INFO: encoder_type : rnn [2021/04/08 16:41:30] root INFO: hidden_size : 96 [2021/04/08 16:41:30] root INFO: name : SequenceEncoder [2021/04/08 16:41:30] root INFO: Transform : None [2021/04/08 16:41:30] root INFO: algorithm : CRNN [2021/04/08 16:41:30] root INFO: model_type : rec [2021/04/08 16:41:30] root INFO: Eval : [2021/04/08 16:41:30] root INFO: dataset : [2021/04/08 16:41:30] root INFO: data_dir : ../train_data/ic15_data/ [2021/04/08 16:41:30] root INFO: label_file_list : ['../train_data/ic15_data/rec_gt_test.txt'] [2021/04/08 16:41:30] root INFO: name : SimpleDataSet [2021/04/08 16:41:30] root INFO: transforms : [2021/04/08 16:41:30] root INFO: DecodeImage : [2021/04/08 16:41:30] root INFO: channel_first : False [2021/04/08 16:41:30] root INFO: img_mode : BGR [2021/04/08 16:41:30] root INFO: CTCLabelEncode : None [2021/04/08 16:41:30] root INFO: RecResizeImg : [2021/04/08 16:41:30] root INFO: image_shape : [3, 32, 100] [2021/04/08 16:41:30] root INFO: KeepKeys : [2021/04/08 16:41:30] root INFO: keep_keys : ['image', 'label', 'length'] [2021/04/08 16:41:30] root INFO: loader : [2021/04/08 16:41:30] root INFO: batch_size_per_card : 1 [2021/04/08 16:41:30] root INFO: drop_last : False [2021/04/08 16:41:30] root INFO: num_workers : 4 [2021/04/08 16:41:30] root INFO: shuffle : False [2021/04/08 16:41:30] root INFO: use_shared_memory : False [2021/04/08 16:41:30] root INFO: Global : [2021/04/08 16:41:30] root INFO: cal_metric_during_train : True [2021/04/08 16:41:30] root INFO: character_dict_path : ../ppocr/utils/dict/innovalor_dict.txt [2021/04/08 16:41:30] root INFO: character_type : ch [2021/04/08 16:41:30] root INFO: checkpoints : None [2021/04/08 16:41:30] root INFO: debug : False [2021/04/08 16:41:30] root INFO: distributed : False [2021/04/08 16:41:30] root INFO: epoch_num : 10 [2021/04/08 16:41:30] root INFO: eval_batch_step : [0, 10] [2021/04/08 16:41:30] root INFO: infer_img : ../doc/imgs_words_en/word_10.png [2021/04/08 16:41:30] root INFO: infer_mode : False [2021/04/08 16:41:30] root INFO: log_smooth_window : 2 [2021/04/08 16:41:30] root INFO: max_text_length : 25 [2021/04/08 16:41:30] root INFO: pretrained_model : ../output/rec/ic15/best_accuracy [2021/04/08 16:41:30] root INFO: print_batch_step : 1 [2021/04/08 16:41:30] root INFO: save_epoch_step : 3 [2021/04/08 16:41:30] root INFO: save_inference_dir : None [2021/04/08 16:41:30] root INFO: save_model_dir : ../output/rec/ic15/ [2021/04/08 16:41:30] root INFO: use_gpu : True [2021/04/08 16:41:30] root INFO: use_space_char : True [2021/04/08 16:41:30] root INFO: use_visualdl : False [2021/04/08 16:41:30] root INFO: Loss : [2021/04/08 16:41:30] root INFO: name : CTCLoss [2021/04/08 16:41:30] root INFO: Metric : [2021/04/08 16:41:30] root INFO: main_indicator : acc [2021/04/08 16:41:30] root INFO: name : RecMetric [2021/04/08 16:41:30] root INFO: Optimizer : [2021/04/08 16:41:30] root INFO: beta1 : 0.9 [2021/04/08 16:41:30] root INFO: beta2 : 0.999 [2021/04/08 16:41:30] root INFO: lr : [2021/04/08 16:41:30] root INFO: learning_rate : 0.0005 [2021/04/08 16:41:30] root INFO: name : Adam [2021/04/08 16:41:30] root INFO: regularizer : [2021/04/08 16:41:30] root INFO: factor : 0 [2021/04/08 16:41:30] root INFO: name : L2 [2021/04/08 16:41:30] root INFO: PostProcess : [2021/04/08 16:41:30] root INFO: name : CTCLabelDecode [2021/04/08 16:41:30] root INFO: Train : [2021/04/08 16:41:30] root INFO: dataset : [2021/04/08 16:41:30] root INFO: data_dir : ../train_data/ic15_data/ [2021/04/08 16:41:30] root INFO: label_file_list : ['../train_data/ic15_data/rec_gt_train.txt'] [2021/04/08 16:41:30] root INFO: name : SimpleDataSet [2021/04/08 16:41:30] root INFO: transforms : [2021/04/08 16:41:30] root INFO: DecodeImage : [2021/04/08 16:41:30] root INFO: channel_first : False [2021/04/08 16:41:30] root INFO: img_mode : BGR [2021/04/08 16:41:30] root INFO: CTCLabelEncode : None [2021/04/08 16:41:30] root INFO: RecResizeImg : [2021/04/08 16:41:30] root INFO: image_shape : [3, 32, 100] [2021/04/08 16:41:30] root INFO: KeepKeys : [2021/04/08 16:41:30] root INFO: keep_keys : ['image', 'label', 'length'] [2021/04/08 16:41:30] root INFO: loader : [2021/04/08 16:41:30] root INFO: batch_size_per_card : 1 [2021/04/08 16:41:30] root INFO: drop_last : True [2021/04/08 16:41:30] root INFO: num_workers : 8 [2021/04/08 16:41:30] root INFO: shuffle : True [2021/04/08 16:41:30] root INFO: use_shared_memory : False [2021/04/08 16:41:30] root INFO: train with paddle 2.0.0 and device CUDAPlace(0) [2021/04/08 16:41:30] root INFO: Initialize indexs of datasets:['../train_data/ic15_data/rec_gt_train.txt'] [2021/04/08 16:41:30] root INFO: Initialize indexs of datasets:['../train_data/ic15_data/rec_gt_test.txt'] W0408 16:41:30.284492 13440 device_context.cc:362] Please NOTE: device: 0, GPU Compute Capability: 7.5, Driver API Version: 11.1, Runtime API Version: 10.2 W0408 16:41:30.287503 13440 device_context.cc:372] device: 0, cuDNN Version: 7.6. [2021/04/08 16:41:32] root INFO: load pretrained model from ['../output/rec/ic15/best_accuracy'] [2021/04/08 16:41:32] root INFO: train dataloader has 3 iters, valid dataloader has 3 iters [2021/04/08 16:41:32] root INFO: During the training process, after the 0th iteration, an evaluation is run every 10 iterations [2021/04/08 16:41:32] root INFO: Initialize indexs of datasets:['../train_data/ic15_data/rec_gt_train.txt'] [2021/04/08 16:41:32] root INFO: epoch: [1/10], iter: 1, lr: 0.000500, loss: 200.846405, acc: 0.000000, norm_edit_dis: 0.000000, reader_cost: 0.01001 s, batch_cost: 0.20338 s, samples: 2, ips: 9.83362 [2021/04/08 16:41:32] root INFO: epoch: [1/10], iter: 2, lr: 0.000500, loss: 199.678284, acc: 0.000000, norm_edit_dis: 0.000000, reader_cost: 0.00000 s, batch_cost: 0.05660 s, samples: 1, ips: 17.66886

Process finished with exit code -1073741819 (0xC0000005)

Can someone help me, please?

MissPenguin commented 3 years ago

6 images are too few

DenisaLicu commented 3 years ago

It was a Windows problem, running on Linux works