OPENCV_DIR= '/home/mhill/projects/alpr/libraries/opencv/bin'
the readme file asks me to change the above mentioned path but i used the easy way to install openalpr on ubuntu and i can't find the opencv path with this folder :SAMPLE_CREATOR = OPENCV_DIR + '/opencv_createsamples'. please help. @matthill

Hello,

I have been messing around with openalpr project for somedays now and been reading about it as much as i could however theres a problem which remains unanswered.

Im trying to train openalpr plate detector for certain angles, i gathered around 1000 plates as positive image changed the prep script regarding to my needs and executed the script, for some reason i ran into the problem which is written below :

TRAIN
Traceback (most recent call last):
  File "./prep.py", line 184, in <module>
    execStr = '%s/opencv_traincascade %s %s %s %s -numPos %d -numNeg %d -maxFalseAlarmRate 0.45 -featureType LBP -numStages 13' % (OPENCV_DIR, data_arg, vector_arg, bg_arg, width_height_arg, num_pos_samples, num_neg_samples )
NameError: name 'num_neg_samples' is not defined

somehow after executing /prep.py train command variable "num_neg_samples" can not be located in python, after a quick look i added
num_neg_samples = file_len(NEGATIVE_INFO_FILE)
below " num_pos_samples = file_len(POSITIVE_INFO_FILE)" to the line 173 this took care of my variable problem but after executing train command process stops in stage 2

===== TRAINING 2-stage =====
<BEGIN
POS count : consumed   1 : 1
Train dataset for temp stage can not be filled. Branch training terminated.

thanks in advance :)

hello, can you please send me an example on the arguments used on crop_plates.py for full function

thank you

.

Hello,

OpenCV opencv_traincascade is monocore out of the box. If you compile with cmake -DWITH_TBB=1 then it builds with multiple core support, BUT it's still not enough because code in traincascade isn't parallelized.

Here is the discussion about traincascade limitations:

http://answers.opencv.org/question/63/enable-multithreading-with-tbb-during-cascade/

I read something about installing IPP to solve it, I'll give it a try.

Someone help me to fix this issue? I have a 4000 neg pictures and 1000 positives pictures. When I run that it gives me that error.

I don't know where I should fix this. Help me. TIA

Hello,
I'm working on training Australian License plates.

I have used 180 Australian license plate positive image and made numStages to 12.
After finishing all the tutorial steps, I didn't find the .xml file "not generated"!!
Please, Can you help me?

Best Regards,
Karem Elzftawy

In the prep.py file for training for country, the dimensions of width and height are in which units, while using this command
python crop_plates.py --input_dir= --out_dir= --plate_width=190 --plate_height=46(mentioned in mm)

Hi,
I've several error running the the python prep.py script on windows 7 using python 2.7.9.
error 1, "from PIL import Image" => the module name incorrect.
changed it from "from pil import Image" to "import image", but still the module Image is unknown.

error 2 is related to directory paths.
original path in script was: "/home/mhill/projects/alpr/samples/training/"
I changed it to: "C:\trainer" and I get error when the script concatenates it in the following instruction:
INPUT_NEGATIVE_DIR = BASE_DIR + 'raw-neg/'
by giving this wrong result: "C:\trainerraw-neg/."
I changed it to: "C:\trainer" and the error is: syntax error: EOL while scanning string literal

seems like the script is written for unix-based system.
can anyone please help?

I found the images in raw-negative have nothing to do with the vehicle license plate. Should I just use it or have to create raw-negative images by myself? Thanks.

i successfully generated the cascade.xml file , rename it to eu2 and placed it in proper folder but using python i am unable to load the eu2.xml file , while .xml files where are present by default are loaded successfully and are working correctly..
i am using windows and opencv3. while openALPR versions is 3.3.1

I reduced the number of pos and neg images but still get this error.
Does anyone know the solution? I google it and most solutions suggest reduce neg and pos, but it didn't work.

Here is the command line I used for training:

/usr/local/Cellar/opencv/3.4.1_5/bin/opencv_traincascade -data /Users/xxx/Desktop/new-ocr/out// -vec /Users/xxx/Desktop/new-ocr/positive/vecfile.vec -bg /Users/xxx/Desktop/new-ocr/negative/negative.txt -w 70 -h 39 -numPos 640 -numNeg 4705 -maxFalseAlarmRate 0.45 -featureType LBP -numStages 11

But the train stopped at Stage 7:

===== TRAINING 7-stage =====
<BEGIN
POS count : consumed   640 : 650
NEG count : acceptanceRatio    0 : 0
Required leaf false alarm rate achieved. Branch training terminated.

What could be the reason? Thanks.

Hello,

I'm facing a probem when using this training method :

exp of error : SyntaxError: Missing parentheses in call to 'print'. Did you mean print(execStr)?

My question is : the script isn't updated to work with Python 3.X ?

Thanks.

Best Regards,

For adding new region(ua), i do:

1. Collecting files with license plate and crop it (with image clipper).
2. Place cropping files into /usr/local/src/train-detector/ua
3. Doing positives:
   root@ubuntu:/usr/local/src/train-detector# ./prep.py pos
   Pos
   /usr/bin/opencv_createsamples -vec /usr/local/src/train-detector/positive/vecfile.vec -w 52 -h 13 -info /usr/local/src/train-detector/positive/positive.txt -num 20
   Info file name: /usr/local/src/train-detector/positive/positive.txt
   Img file name: (NULL)
   Vec file name: /usr/local/src/train-detector/positive/vecfile.vec
   BG file name: (NULL)
   Num: 20
   BG color: 0
   BG threshold: 80
   Invert: FALSE
   Max intensity deviation: 40
   Max x angle: 1.1
   Max y angle: 1.1
   Max z angle: 0.5
   Show samples: FALSE
   Width: 52
   Height: 13
   Create training samples from images collection...
   Done. Created 20 samples
4. Doing negatives:
   output only -
   root@ubuntu:/usr/local/src/train-detector# ./prep.py neg
   Neg
5. In this case the train fold down with error
   root@ubuntu:/usr/local/src/train-detector# ./prep.py train
   TRAIN
   Traceback (most recent call last):
   File "./prep.py", line 170, in
   num_neg_samples = file_len(NEGATIVE_INFO_FILE)
   File "./prep.py", line 54, in file_len
   return i + 1

Why negatives does not create.

Hey folks,

please delete my message if this is the wrong place to ask a question. I am working or VELUX, a company that build roof windows. Since many decades our windows have a type plate (see attached picture). I would like to create an application that is able to read the data from the type plates. Is OpenALPR the right way to approach taht?

Cheers,
Klaus

Hi
I am trying to train the detector using the python script. the neg, pos is working but as soon as i running the train code, windows show me this error

Can anyone tell me what is running wrong and what should I do to fix it ?

Thank you

Hi, I'm using OSX 10.10.2 (brew)

The stages seems that are not moving forward
What is missing? Not error messages and the files already are in the respective folders

Here is what I did using the folder "eu" as input..

/usr/local/Cellar/opencv/2.4.11_1/bin/opencv_traincascade -data /train-detector/out/ -vec /train-detector/positive/vecfile.vec -bg /train-detector/negative/negative.txt -w 52 -h 13 -numPos 1347 -numNeg 3248 -featureType LBP -numStages 20
PARAMETERS:
cascadeDirName: /train-detector/out/
vecFileName: /train-detector/positive/vecfile.vec
bgFileName: /train-detector/negative/negative.txt
numPos: 1347
numNeg: 3248
numStages: 20
precalcValBufSize[Mb] : 256
precalcIdxBufSize[Mb] : 256
stageType: BOOST
featureType: LBP
sampleWidth: 52
sampleHeight: 13
boostType: GAB
minHitRate: 0.995
maxFalseAlarmRate: 0.5
weightTrimRate: 0.95
maxDepth: 1
maxWeakCount: 100

===== TRAINING 0-stage =====
<BEGIN
POS count : consumed 1347 : 1347
NEG count : acceptanceRatio 3248 : 1

Here is where not continues..

Got the following error messgae:

No module named cv

Any idea? Thanks.

In crop_plates.py I saw that the parameter for that was a pixels, how about the prep.py?

What was the exact value that asking for height and width in prep.py?
is that the char height and width of a plate number?

or the size of the full plate number?

Please anyone help me in this issue.

So i wanted to ask , after tagging the plates in the tagger tool , i got simple yaml file , like this :
image_file: 1.jpeg image_width: 640 image_height: 480 plate_corners_gt: 453 322 547 315 547 338 453 347 region_code_gt: mk plate_number_gt: SK2235AR plate_inverted_gt: false
Now how can i use this ? also when i try to run the crop plates script , it says 0 images found in the output folder , even if i add some images it still says 0 , what am i doing wrong ? can someone please walk me through ? or at least little more explanation
Thank you

what kind of images should be inside negative sample folder to rich higher accuracy?
should I put any image(whiteout license plate ) with any size inside that or it's better to resize them as same as my positive samples are?
what about the color? should I use grayscale images or RGB images give me higher accuracy?

Traceback (most recent call last):
File "prep.py", line 77, in
for neg_file in os.listdir(INPUT_NEGATIVE_DIR):
OSError: [Errno 2] No such file or directory: '/home/mhill/projects/alpr/samples/training/raw-neg/'

I have tried to train the detector with lots of images (jpeg). But I allways get:

===== TRAINING 0-stage =====
<BEGIN
POSKillednt samples: 18957

or

===== TRAINING 0-stage =====
<BEGIN
POS cKilled samples: 18919

Have I make something wrong? What can I do to get the .xml?

Can anyone please tell me what would be causing the below crash.
Thanks

I run the following command, but nothing happened.

./prep.py neg

It only shows: Neg. Then nothing happened. May I know what should I include in raw-neg folder?

But when I run:

./prep.py pos

Pos
/usr/local/Cellar/opencv/3.4.1_5/bin/opencv_createsamples -vec /Users/xxx/Projects/ocr/train-detector/positive/vecfile.vec -w 70 -h 50 -info /Users/xxx/Projects/ocr/train-detector/positive/positive.txt -num 1
Info file name: /Users/xxx/Projects/ocr/train-detector/positive/positive.txt
Img file name: (NULL)
Vec file name: /Users/xxx/Projects/ocr/train-detector/positive/vecfile.vec
BG  file name: (NULL)
Num: 1
BG color: 0
BG threshold: 80
Invert: FALSE
Max intensity deviation: 40
Max x angle: 1.1
Max y angle: 1.1
Max z angle: 0.5
Show samples: FALSE
Width: 70
Height: 50
Max Scale: -1
RNG Seed: 12345
Create training samples from images collection...
Done. Created 1 samples

Only 1 samples created. Is this correct?

when using OpenALPR Ionic with Cordoava plugin for some Indian Number Plates data unable to getting .how to achive this one .i am getting EU, US data geting but Indian Number plate unable getting .

Hello,

Based on a discussion on the forum, opencv_traincascade often fails because you have to guess the number of samples and pass numPos about 90% of the sample for it to work.

The error is pretty cryptic like "Can not get new positive sample. The most possible
reason is insufficient count of samples in given vec-file."

Here are links about how to tackle this:

http://stackoverflow.com/questions/10863560/haar-training-opencv-assertion-failed
http://code.opencv.org/issues/1834

Just creating this issue for others and maybe a PR will eventually come.

Hi,
OS: windows 10
I am trying to train for Indian number plates, and I have some image in positive directory.
indian license plate standard is 500 * 120 mm, so I created a 500 * 120 mm image and in pixels it converts to 1417*340 px.

I run the 'prep.py neg' and it counts all files as it creates a negative.txt in negative directory and there are all negative files are listed.

I tried running 'prep.py pos' and there was 40 images in my folder 'in'(country code) and positive directory was empty before running the command and after executing the 'prep.py pos' command all files got copied from 'in' folder to 'positive' folder and output of the command is below.

C:\Users\LaserTrac\Downloads\alpr\train-detector-master>python prep.py pos
Pos
Exception reading image file: positive.txt
C:/Users/LaserTrac/Downloads/opencv1/opencv/build/x64/vc14/bin/opencv_createsamples -vec C:/Users/LaserTrac/Downloads/alpr/train-detector-master/positive/vecfile.vec -w 1417 -h 340 -info C:/Users/LaserTrac/Downloads/alpr/train-detector-master/positive/positive.txt -num 0
Info file name: C:/Users/LaserTrac/Downloads/alpr/train-detector-master/positive/positive.txt
Img file name: (NULL)
Vec file name: C:/Users/LaserTrac/Downloads/alpr/train-detector-master/positive/vecfile.vec
BG  file name: (NULL)
Num: 0
BG color: 0
BG threshold: 80
Invert: FALSE
Max intensity deviation: 40
Max x angle: 1.1
Max y angle: 1.1
Max z angle: 0.5
Show samples: FALSE
Width: 1417
Height: 340
Max Scale: -1
RNG Seed: 12345
Create training samples from images collection...
Done. Created 0 samples

and positive.txt file was empty.
please help.

File "crop_plates.py", line 34
print "input_dir (%s) doesn't exist"
^
SyntaxError: invalid syntax

what to do if this happens?

hi
I have a question.

WIDTH and HEIGHT is mm or px ?

for example :

WIDTH=52

HEIGHT=13

COUNTRY='eu'

but Real HEIGHT size is 11 mm.