Extract Tables from Image Documents | Paddle Paddle | Paddleocr | OCR | Text Extraction |

[@MitvHome-x8y]

Thank you so much . I do the same as you but I always get an empty Excel. Why would it be?

[@hemu2723]

Hey, have you got the mistake?

[@ram_rahim_creations_officials]

Hi @karndeep Thank you for sharing. Will it work if my table doesn't have vertical and horizontal lines?

[@ShreyasG-d2n]

it should

[@nomuchohan]

Dude, please explain how to use the PPstructure from paddle paddle into our own custom code

[@ajithn7336]

I tried and i always get an empty excel only.

[@xy4611]

same

[@niroshiniedayaratne4066]

My output is always empty xlsx file. What could be the reason? Thanks in advance!

[@karndeepsingh]

May be OCR is unable to read the table content

[@kishoripawar2522]

@@karndeepsingh Is there any prerequisite for input image? Like resolution more than X or something like that? Because for me as well, output is empty.

[@kishoripawar2522]

@@karndeepsingh Eve with high resolution image output is empty, when I checked show.html, the blue box is not able to correctly locate the table in image. So I think as there is no text inside blue box, there is empty csv. Please correct me if I am wrong.

[@pavitrabiradar6334]

@@kishoripawar2522 even iam getting empty xlsx as output did you found any solution?

[@보라색사과-l1r]

any update for this issue? I am facing this issue after trying another ocr model... please help

[@avikalchauhan9907]

when I am running the code predict_table.py file is not there

[@kiddicode6897]

How can I apply Google Vision after table is recognized?

[@venkatesanr9455]

Thanks for the great explanation and video. I have some doubts like 1. Is paddleocr is open source library and anyone can use? 2.Whether we can fine-tune ocr models like easyocr, paddleocr libraries, Kindly reply and share links that will be useful for reading/learning purpose. 3. Whether huggingface library has ocr models?

[@karndeepsingh]

1. Yes, paddlepaddle is an open source library. 2. You can train OCR model using paddleocr 3. Huggingface may not have OCR models.

[@venkatesanr9455]

@@karndeepsingh Thanks for your kind replies.Can you share any links for finetuning models of easyocr/paddleocr( I hav searched for easyocr but not obtained proper links for finetuning tasks)

[@karndeepsingh]

@@venkatesanr9455 you can check paddleocr github for the same.

[@venkatesanr9455]

@@karndeepsingh Ok thanks a lot

[@NickWindham]

@@venkatesanr9455 Watch his video titled OCR Text from PDFs and Image Documents using docTR | Better than Tesseract OCR | Text Extraction

[@ganeshrajv130]

wont this support long image table

[@eliaweiss1]

Thanks, all I get is empty cells

[@jayeshnikam3279]

This is kind of urgent. What if on some page half of the table is in one page and other half is on 2nd page. What can be done on such situation? Will the model recognize it??. i highly expect your answer as I am currently working on it. Thank you! :)

[@karndeepsingh]

In such situations, you need to search identifier in the page that consider that half of the information in going to next page. Model can only help you extract or detect table but on top of that you need to apply logic to know whether its full information or half information

[@poojabhandari631]

getting this error error: legacy-install-failure × Encountered error while trying to install package. ╰─> PyMuPDF what to do??

[@Smddlvvs]

How to make this code work on pdf files with multiple pages

[@karndeepsingh]

Pass each page of PDF to the model

[@Smddlvvs]

@@karndeepsingh i have tried but i am unable to iterate

[@texasfossilguy]

you need to write code to iterate each page of it. Ask chatgpt or google that, ive seen it. If I find it Ill let you know.

[@Smddlvvs]

@@texasfossilguy yaaaa pls let me know if you find one

[@AliAlias]

Use other python libraries to extract pdf to images then ocr it one by one using loop 😊

[@louieelumbaring1790]

how did you get the vqa folder? Sorry I was trying to do all the steps you did and find error on the last line, i have no idea to fix it. Thanks in advance! [Errno 2] No such file or directory: 'PaddleOCR/ppstructure' /content/PaddleOCR/ppstructure/inference Traceback (most recent call last): File "/content/PaddleOCR/ppstructure/table/predict_table.py", line 230, in main(args) File "/content/PaddleOCR/ppstructure/table/predict_table.py", line 149, in main image_file_list = get_image_file_list(args.image_dir) File "/content/PaddleOCR/ppocr/utils/utility.py", line 60, in get_image_file_list raise Exception("not found any img file in {}".format(img_file)) Exception: not found any img file in /content/PaddleOCR/ppstructure/table/image1.png

[@rivamalik9575]

Provide absolute path to the image that is placed in drive. For example /content/gdrive/MyDrive/PaddleOCR/ppstructure/table/image1.png and also ensure that the image is place in the table folder that you have mentioned in the exception statement.

[@pavitrabiradar6334]

Hello Iam always getting output as empty xlsx file could you please help me here.

[@karndeepsingh]

May be OCR is not working that great. You may consider replacing OCR.

[@ShivShankarDutta1]

getting this error executing #%cd PaddleOCR/ppstructure !python3 /content/PaddleOCR/ppstructure/table/predict_table.py --det_model_dir=inference/en_PP-OCRv3_det_infer --rec_model_dir=inference/en_ppocr_mobile_v2.0_table_rec_infer --table_model_dir=inference/en_ppocr_mobile_v2.0_table_structure_infer --image_dir=/content/PaddleOCR/ppstructure/table_2.png --rec_char_dict_path=../ppocr/utils/dict/table_dict.txt --table_char_dict_path=../ppocr/utils/dict/table_structure_dict.txt --det_limit_side_len=736 --det_limit_type=min --output ./output/table Traceback (most recent call last): File "/content/PaddleOCR/ppstructure/table/predict_table.py", line 30, in import tools.infer.predict_det as predict_det File "/content/PaddleOCR/tools/infer/predict_det.py", line 31, in from ppocr.data import create_operators, transform File "/content/PaddleOCR/ppocr/data/__init__.py", line 35, in from ppocr.data.imaug import transform, create_operators File "/content/PaddleOCR/ppocr/data/imaug/__init__.py", line 47, in from .ct_process import * File "/content/PaddleOCR/ppocr/data/imaug/ct_process.py", line 22, in import Polygon as plg ModuleNotFoundError: No module named 'Polygon'

[@rohithuria1168]

how to fix this error ?

[@goswamidivyang2010]

@@rohithuria1168 Did you get any fix for that? I am also facing the same error

[@luisvite3766]

Me too

[@xy4611]

maybe !pip install polygon

[@shobhitsadwal6081]

it is not working for me .

[@rajeshroyal5922]

i have tried with vs code and colab but iam getting this error python3: can't open file '/PaddleOCR/ppstructure/table/predict_table.py': [Errno 2] No such file or directory

[@thepresistence5935]

change the path bro

[@rajeshroyal5922]

@@thepresistence5935 I tried with change of path also getting same error

[@thepresistence5935]

@@rajeshroyal5922 It's working fine for me, put quotes.

[@vogel2499]

I suspect text ocr is independent from table detection/recognition. You could replaced it with easyocr/pytesseract without ruining the structure.

[@shwetabhilare9473]

[Errno 2] No such file or directory: 'PaddleOCR/ppstructure' /content/PaddleOCR/ppstructure/inference Traceback (most recent call last): File "/content/PaddleOCR/ppstructure/table/predict_table.py", line 30, in import tools.infer.predict_det as predict_det File "/content/PaddleOCR/tools/infer/predict_det.py", line 31, in from ppocr.data import create_operators, transform File "/content/PaddleOCR/ppocr/data/__init__.py", line 35, in from ppocr.data.imaug import transform, create_operators File "/content/PaddleOCR/ppocr/data/imaug/__init__.py", line 47, in from .ct_process import * File "/content/PaddleOCR/ppocr/data/imaug/ct_process.py", line 22, in import Polygon as plg ModuleNotFoundError: No module named 'Polygon' getting this error please help.

[@madhavkumarpancholi9842]

get to the point dude.

[@anouaraadoud58]

Errno 2] No such file or directory: 'PaddleOCR/ppstructure' /content/PaddleOCR/ppstructure/inference Traceback (most recent call last): File "/content/PaddleOCR/ppstructure/table/predict_table.py", line 230, in main(args) File "/content/PaddleOCR/ppstructure/table/predict_table.py", line 153, in main table_sys = TableSystem(args) File "/content/PaddleOCR/ppstructure/table/predict_table.py", line 67, in __init__ self.text_detector = predict_det.TextDetector(copy.deepcopy( File "/content/PaddleOCR/tools/infer/predict_det.py", line 141, in __init__ self.predictor, self.input_tensor, self.output_tensors, self.config = utility.create_predictor( File "/content/PaddleOCR/tools/infer/utility.py", line 199, in create_predictor raise ValueError( ValueError: not find model.pdmodel or inference.pdmodel in inference/en_PP-OCRv3_det_infer

[@SaniyaFarash]

I am getting the same error. please tell how to solve this

[@rajeshroyal5922]

i can't open predict_table.py file getting the same error python3: can't open file '/PaddleOCR/ppstructure/table/predict_table.py': [Errno 2] No such file or directory how can i resolve

[@kiddicode6897]

%cd /content/PaddleOCR: go to the Path !mkdir inference: create folder "inference" inside the Path below "/content/PaddleOCR" %cd /content/PaddleOCR/inference: go to the PATH download and unzip file inside "inference"