Can not train any image. need advice

[Russell]

Hi everyone. 

Using the tesseract to recognize a racing number has been training.

However, an error occurs as follows: If I train any image.

-------------------------------------

FAIL!

APPLY_BOXES: boxfile line 1/7 ((69,48),(111,151)): FAILURE! Couldn't find a matching blob

FAIL!

APPLY_BOXES: boxfile line 2/1 ((129,48),(142,151)): FAILURE! Couldn't find a matching blob

FAIL!

APPLY_BOXES: boxfile line 3/4 ((160,48),(200,151)): FAILURE! Couldn't find a matching blob

FAIL!

APPLY_BOXES: boxfile line 4/5 ((205,48),(245,151)): FAILURE! Couldn't find a matching blob

FAIL!

APPLY_BOXES: boxfile line 5/6 ((248,48),(287,151)): FAILURE! Couldn't find a matching blob

APPLY_BOXES:

   Boxes read from boxfile:       6

   Boxes failed resegmentation:       5

   Found 1 good blobs.

   Leaving 3 unlabelled blobs in 0 words.

-------------------------------------

The following is a result of creating a box file for training the "G71456" from original.jpg file attached.

~ 7 0 300 225 0

I was modified as follows:

G 10 48 60 151 0

7 69 48 111 151 0

1 129 48 142 151 0

4 160 48 200 151 0

5 205 48 245 151 0

6 248 48 287 151 0

Then, the above error when creating a tr file has occurred.

Unusual is edit.jpg taken only at the "G71456" from original.jpg can successfully training.

However, the traineddata can not recognize original.jpg.

Why do this?

I do not understand.

Maybe because original.jpg are mixed are various characters?

How to properly train is original.jpg?

Any ideas?

Thanks!