i was facing same issue but solve.....

[Brutal15229]

          i was facing same issue but solve.....

label_file_list= ratio_list

Originally posted by @jitesh-rathod in https://github.com/PaddlePaddle/PaddleOCR/issues/1996#issuecomment-811715798

[Brutal15229]

when i am doing this File "/data/data/aditya_llama/Ocr_training/archive/PaddleOCR/ppocr/data/_init.py", line 107, in build_dataloader dataset = eval(module_name)(config, mode, logger, seed) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/data/data/aditya_llama/Ocr_training/archive/PaddleOCR/ppocr/data/simple_dataset.py", line 56, in _init self.data_lines = self.get_image_info_list(label_file_list, ratio_list) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/data/data/aditya_llama/Ocr_training/archive/PaddleOCR/ppocr/data/simple_dataset.py", line 73, in get_image_info_list lines = random.sample(lines, round(len(lines) * ratio_list[idx])) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ TypeError: type str doesn't define round method

[Brutal15229]

pls help me i stuck over here

[GreatV]

It seems you are referencing a PaddleOCR-related issue where the user claimed to have resolved their problem with the following code:

label_file_list = ratio_list

The context of your question seems to suggest that you are encountering similar issues with PaddleOCR, particularly errors like KeyError: 'label' or other data-related discrepancies during processing.

---

Potential Causes and Solutions:

1. Error: KeyError: 'label'

   • Cause: This error typically occurs when the data loader or transformation pipeline is unable to find the label key in the input data or annotation file. This might happen due to:

     • Improperly formatted dataset annotations.
     • Missing label fields in the dataset JSON or CSV files.
     • Incorrect transforms or keys being used in the configuration.

   • Solution:

     • Check your dataset annotation files to ensure that each entry contains a valid label key. For example:

       {
       "transcription": "example text",
       "points": [[x1, y1], [x2, y2], [x3, y3], [x4, y4]],
       "label": "class_name"
       }

       If the label field is missing, you need to add it manually or modify your dataset generation script to include it.

     • Ensure that your KeepKeys transform in the configuration includes label if it's being used in the pipeline:

       KeepKeys:
       keep_keys: ['image', 'label', 'shape', ...]

     • Update the VQATokenLabelEncode or other encoding transforms to correctly map the label field if required.

2. Misalignment of label_file_list and ratio_list

   • Cause: During training, label_file_list and ratio_list are used together to load and balance datasets. If these lists are not aligned in length or content, errors can occur.
   • Solution:
     • Ensure that the label_file_list in your configuration corresponds to the correct file paths of your dataset annotations.
     • Verify that ratio_list contains the same number of entries as label_file_list and that the ratios are correctly set (e.g., [1.0] for a single dataset or [0.7, 0.3] for train/validation splits).

3. Error in table/predict_table.py

   • Cause: If you are working with table recognition, errors can occur due to improperly structured input data or configuration mismatches.
   • Solution:
     • Ensure that the table annotation files are in the correct format required by PaddleOCR.
     • If the error occurs during prediction, check the predict_table.py script and ensure that the input image paths and preprocessing steps are configured correctly.

4. General Debugging Approach:

   • Enable debug logs by setting show_log=True in the PaddleOCR configuration or script.
   • Run the script with a smaller subset of your dataset to isolate problematic entries.
   • Validate your data pipeline step by step (e.g., loading images, checking annotations, running transforms).

---

Key Resources:

• Relevant Issues and Discussions:

  • 表格识别 报错 IndexError: list index out of range
  • 多线程识别报错
  • Shape_list error while learning Paddle OCR

• Suggestions from the User's Post: The snippet label_file_list = ratio_list might imply that the user manually assigned the label_file_list variable to resolve an alignment or configuration mismatch. Ensure that both variables are correctly defined in your configuration YAML file.

---

If you can provide more details about your specific issue (e.g., errors, configurations, or dataset structure), I can provide a more targeted solution!

---

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