How to get clean output from easyocr

Question

This is from a opencv+easyocr number plate recognition script opencv crops the image to number plate and gives,clean great output to the easyocr. but what are these bunch of numbers its reading

result = reader.readtext(opencv(mypath))

Result: [([[0, 0], [163, 0], [163, 31], [0, 31]], 'SPHJ > 3764', 0.5565279612963627)]

I know I can get a clean output by this but the thing is it differs from a picture to picture. is there any way only to get the number plate

result = result[0][-2]

Result: SPHJ > 3764

buran · Accepted Answer · 2021-10-17T18:54:03.070

2

As stated in the docs:

The output will be in a list format, each item represents a bounding box, the text detected and confident level, respectively.

It's the coordinates of the box where the text is located. And the last item is the confidence level.

[[0, 0], [163, 0], [163, 31], [0, 31]] -> the coordinates of the 4 corners
'SPHJ > 3764' -> the text
0.5565279612963627 -> confidence level

Just use result[0][1] to get the text. Note, result may have multiple text boxes detect, so you need to access by index or iterate over it.

for item in result:
    print(item[1])

edited Oct 17 '21 at 18:54

answered Oct 17 '21 at 16:50

buran

13,682
10
36
61

so i need to run a loop to result[0][1]->result[1][1] like that right but how do i know that "if" the output have got 2 or more locations – Vishwa Shehan Oct 17 '21 at 18:51
I edited my answer to include sample of the loop. – buran Oct 17 '21 at 18:54

score 1 · Answer 2 · answered Apr 03 '23 at 21:23

I think the question is to get "pure" texts without position and confidence, then simply add 'detail=0' to readtext()

result = reader.readtext(opencv(mypath), detail=0)
# output: result = ['SPHJ > 3764'] or multiple items if the image contains more texts

How to get clean output from easyocr

2 Answers2