PyTesseract image_to_data function isn't recognizing my image

Question

I'm using pytesseract to return the coordinates of the objects in an image.

By using this piece of code:

import pytesseract
from pytesseract import Output
import cv2
img = cv2.imread('wine.jpg')

d = pytesseract.image_to_data(img, output_type=Output.DICT)
    print(d)

for i in range(n_boxes):
    (x, y, w, h) = (d['left'][i], d['top'][i], d['width'][i], d['height'][i])
    cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)

cv2.imshow('img', img)
cv2.waitKey(0)

I get that:

{'level': [1, 2, 3, 4, 5, 5, 2, 3, 4, 5, 4, 5, 2, 3, 4, 5], 'page_num': [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], 'block_num': [0, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3], 'par_num': [0, 0, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 1], 'line_num': [0, 0, 0, 1, 1, 1, 0, 0, 1, 1, 2, 2, 0, 0, 1, 1], 'word_num': [0, 0, 0, 0, 1, 2, 0, 0, 0, 1, 0, 1, 0, 0, 0, 1], 'left': [0, 485, 485, 485, 485, 612, 537, 537, 555, 555, 537, 537, 454, 454, 454, 454], 'top': [0, 323, 323, 323, 323, 324, 400, 400, 400, 400, 426, 426, 0, 0, 0, 0], 'width': [1200, 229, 229, 229, 115, 102, 123, 123, 89, 89, 123, 123, 296, 296, 296, 296], 'height': [900, 29, 29, 29, 28, 28, 40, 40, 15, 15, 14, 14, 892, 892, 892, 892], 'conf': ['-1', '-1', '-1', '-1', 58, 96, '-1', '-1', '-1', 95, '-1', 95, '-1', '-1', '-1', 95], 'text': ['', '', '', '', "JACOB'S", 'CREEK', '', '', '', 'SHIRAZ', '', 'CABERNET', '', '', '', '']}

[image used][ enter image description here ]1

However, when I use this image:

I get that:

{'level': [1, 2, 3, 4, 5], 'page_num': [1, 1, 1, 1, 1], 'block_num': [0, 1, 1, 1, 1], 'par_num': [0, 0, 1, 1, 1], 'line_num': [0, 0, 0, 1, 1], 'word_num': [0, 0, 0, 0, 1], 'left': [0, 0, 0, 0, 0], 'top': [0, 162, 162, 162, 162], 'width': [1200, 0, 0, 0, 0], 'height': [900, 276, 276, 276, 276], 'conf': ['-1', '-1', '-1', '-1', 95], 'text': ['', '', '', '', '']}

Any idea why some image are working and some aren't?

prhmma · Accepted Answer · 2020-05-25 19:53:50Z

1

It is mainly caused by different quality and contrast. it is much easier for the OCR engine to detect texts in desired images. you can add a few pre-processing routines, including thresholding, blurring, histogram equalization and lots of other techniques. it is mainly subjective so I can not provide you with working code, it is more like trial and error to find the best technique for your scope

UPDATE: here is a code that might help you

def preprocessing_typing_detection(inputImage):
    inputImage= cv2.cvtColor(inputImage, cv2.COLOR_BGR2GRAY)
    inputImage= cv2.Laplacian(inputImage, cv2.CV_8U)
    return inputImage

edited May 25, 2020 at 19:53

answered Feb 1, 2020 at 15:40

prhmma

96312 silver badges20 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

rafasalo Over a year ago

Could you provide just a code example? For example, if I want to do a thresholding.

Collectives™ on Stack Overflow

PyTesseract image_to_data function isn't recognizing my image

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related