Ask Your Question
1

Training Neumann Matas classifiers

asked 2015-01-30 00:10:50 -0500

chase gravatar image

I'm trying to train the classifiers in the scene text detection (http://docs.opencv.org/trunk/modules/...) on my own training sets instead of using the default classifiers: trained_classifierNM1.xml and trained_classifierNM2.xml. Any guidance on how to do so would be greatly appreciated. There doesn't seem to be any mention of training on the docs page (or in the code).

I looked at other examples on the web for training classifiers in open CV (http://note.sonots.com/SciSoftware/ha...), and I think where I'm stumbling is on how to define the weak classifiers. Any help or pointers would be greatly appreciated.

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
4

answered 2015-01-30 04:27:47 -0500

Jaykob gravatar image

updated 2017-01-27 04:47:51 -0500

You can use the following code to train your own classifiers: https://github.com/lluisgomez/erfilte... I just did it with success. For this process I modified the JTessBoxEditor application so that recognized characters can be cut out, inverted and exported as individual image files. This can save you much time by avoiding cutting out by hand. If you're interested I would upload it somewhere so that you can use it.

I'm currently trying to improve the grouping phase, because I also need to support single characters which are dismissed at the moment. Therefore I tried to train my own erGrouping classifier with the code on Lluis Gomez's GitHub account, but there are no sample images and my results aren't satisfying so far. So I got back to modifying the exhaustive search algorithm...

Update: I finally uploaded the modded jTessBoxEditor to GitHub. Please see my note in the readme on how to invert the images as this is needed to train the erfilter.

edit flag offensive delete link more

Comments

Hi,I'm also trying to train the classifiers in the scene text detection. I downloaded the erfilter_train and have tried to train, but there are some errors and I don't kown how to solve it. Would you send me your code? My Email is luo.baojuan@163.com. Thanks so much.

Carrot_Luobo gravatar imageCarrot_Luobo ( 2015-02-03 06:34:34 -0500 )edit

Hi, I sent you the code a while ago. Did you get it?

Jaykob gravatar imageJaykob ( 2015-02-24 05:58:08 -0500 )edit

Hi Jaykob, Amazing job! could you send me how did you modified JTessBoxEditor in order to cut out, invert and export the pictures? My email is fran_2_14@hotmail.com Thanks!

Fran Marquez gravatar imageFran Marquez ( 2015-07-26 08:50:39 -0500 )edit

Hi Jaykob, I also try to train my own classifier model with the program from this link https://github.com/lluisgomez/erfilte... but it failed with some error. Can you please send me your code and jTessBoxEditor? Here is my e-mail apichat31@hotmail.com

My bad, Everything is fine once run it in Linux system. Thanks

A.sakarind gravatar imageA.sakarind ( 2015-10-19 06:32:02 -0500 )edit

Hiii, I want to know the procedure to run luiz gomez github code.As i tried it to run on eclipse along with including the opencv libraries but still i am getting some errors...like Field "pixel" could not be resolved etc. what should i do to get out of these errors.

Shaikh Arif gravatar imageShaikh Arif ( 2016-01-19 09:20:29 -0500 )edit

Hi, do you know how to train erGrouping's classifier?

Changxu gravatar imageChangxu ( 2016-04-28 07:19:52 -0500 )edit
4

I know it's very late, but I finally uploaded the jTessBoxEditor to GitHub. Please see my note in the readme on how to invert the images as this is needed to train the erfilter.

Jaykob gravatar imageJaykob ( 2017-01-27 04:46:23 -0500 )edit
Login/Signup to Answer

Question Tools

2 followers

Stats

Asked: 2015-01-30 00:10:50 -0500

Seen: 1,284 times

Last updated: Jan 27 '17