Seven Segments OCR

Configuration
1

Ive set this integration up and I cannot get it to read what I would would be a simple display/

It’s a two digit illuminated 7 segment display. The image in my camera looks like this (from an espcam, its upside down and rotated with ssocr)

camsnap4
camsnap41280×1024 61.9 KB

As per the documentation im trying to get the right settings from the command line. So far Ive got this:

ssocr crop 468 136 165 210 -d -1 rotate 180 shear 25 r_threshold --verbose erosion camsnap4.jpg -D -t 95

This is cropped try to just get one digit. The debug image shows this:

testbild

Which to me looks like a 9, but ssocr keeps telling me its an 8. Its so consisently wrong that Im assuming i must be doing something wrong.

Any ideas?

1 Like
2

In case anyone else ever looks for this I worked out the problem - I had to invert the image. By default it looks for dark text on a light background, mine is illuminated leds so light text on a dark background. adding ‘invert’ to the command line fixed it.

3 Likes
3

Thanks for the solution. I would just like to know where is the image saved as I can’t find it in the config folder?

4

By default it doesn’t save any image but if you add

-D/media/filename.png

To input_arguments: then it will save the debugged output file in your media folder.