Error with det=True

[theophiledebouetduportal]

Detection mode one does not work anymore on the latest version.
System Environment：Ubuntu
Version: paddleocr 2.7.0.3
Command Code：ocr_engine.ocr(filepath, det=True, rec=False)
Complete Error Message：
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()
Fix :
In paddleocr.py :
Ligne 674 " if not dt_boxes" ==> "if not dt_boxes.any():" fix the error.

[bartproo]

same, got this issue too

[theophiledebouetduportal]

import importlib.util
from paddleocr import PaddleOCR

def get_module_path(my_module_name):
    spec = importlib.util.find_spec(my_module_name)
    if spec and spec.origin:
        return spec.origin
    return None

def modify_paddleocr_file(my_file_path):
    with open(my_file_path, "r", encoding="utf-8") as file_to_check:
        lines = file_to_check.readlines()
    # Check if the file has enough lines
    if len(lines) > 674:
        # Modify the specified line
        if "if not dt_boxes" in lines[673]:
            print("content of line 674: ", lines[673])
            lines[673] = "                if not dt_boxes.any():\n"
        else:
            print("The content of line 674 doesn't match the expected content.")
            return
    with open(my_file_path, "w", encoding="utf-8") as file_to_modify:
        file_to_modify.writelines(lines)
    print("File modified successfully!")


module_name = "paddleocr"
file_path = get_module_path(module_name).replace("__init__.py", "paddleocr.py")

if file_path:
    print(f"Found at: {file_path}")
    modify_paddleocr_file(file_path)
else:
    print("Module not found!")


PaddleOCR(use_angle_cls=True, lang="en", ocr_version="PP-OCRv4")
PaddleOCR(use_angle_cls=True, lang="fr", ocr_version="PP-OCRv4")

I have integrated this code to fix the bug in my CD

[george-ghimici]

Hi!
I have the same issue. Any plans to have the fix integrated into paddle distribution ?

[nguyengiaphuc1992]

I have the same issue

[rm-asif-amin]

Hi @theophiledebouetduportal ,
executed this and got "File modified succesfully" on colab, still getting error after this inference code-

from paddleocr import PaddleOCR, draw_ocr
from PIL import Image
from IPython import display

img_path = '/content/image.png'
ocr = PaddleOCR(lang='en')
result = ocr.ocr(img_path,rec=False)
for idx in range(len(result)):
    res = result[idx]
    for line in res:
        print(line)

result = result[0]
image = Image.open(img_path).convert('RGB')
im_show = draw_ocr(image, result, txts=None, scores=None, font_path='/content/PaddleOCR/StyleText/fonts/ko_standard.ttf')
im_show = Image.fromarray(im_show)
im_show.save('result.jpg')
display.Image('result.jpg')

Any suggestion?

[rm-asif-amin]

Solved after making sure updated version was installed-
!pip install "paddleocr>=2.0.1"

And separately, making sure font could be accessed on Colab while drawing visualization of result-

!git clone https://github.com/PaddlePaddle/PaddleOCR

# draw result
from PIL import Image
result = result[0]
image = Image.open(img_path).convert('RGB')
boxes = [line[0] for line in result]
txts = [line[1][0] for line in result]
scores = [line[1][1] for line in result]
im_show = draw_ocr(image, boxes, txts, scores, font_path='/content/PaddleOCR/StyleText/fonts/ko_standard.ttf')
im_show = Image.fromarray(im_show)
im_show.save('result.jpg')

[thinhemb]

at https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.7/paddleocr.py#L681

You can put:

if not dt_boxes:

Replace with:

if len(dt_boxes) == 0:

It will be solved.

[SWHL]

In the branch of main, this bug has been solved.