seeebek / eliteocr Goto Github PK

When I close the application and then try and run it again later it fails to run.

This is on Win 8.1 x64 with the Windows version.

I'm afraid I can't offer any more information than that.

Hi,
ED has gone and added two new commodities, and I can add them (Osmium and SAP 8 Core Container), but what about allowing the rare goods to be scanned and spell-checked?
There are substantial lists of the rare goods that are available around the galaxy, and if you had them recognised, instead of dropped, then wouldn't that be helpful?
I mean, you have the Commodity Editor with a field for "rare" ... why isn't it pre-populated?
Surely the guys and gals around the community who have gone to effort to collate the long lists of rare items would hand them over?
Then, you know that it's an item in the cargo hold if the SELL value is higher than the BUY value.. maybe someone will be interested in the cost/ly increase of certain rare items... and that can be collated by someone... it's just added Sale Price data... they're just commodities with a single source? I know you're only interested in the OCR part of this flow chart, but present that data the same, it'll give the rest the opportunity to take it.

I was thinking to improve EliteOCR, it could store the Galactic Average Price (definable by the user, much like the commodities list) and if it matches a commodity to a price which differs from the GAP by a given percentage (again, defined by the user), it could flag the input for confirmation. This should help with the bad in = bad out.

Can it also clear the discovered station name from the log, as it's not intuitive how the + button operates on reading the station name from the log when you add a new file.

I.e. Currently, you clear all to strip the prices, then may manually blank the station name (because it's not right for what you want to do next), then use the + button to add a new station. Usually this results in the old station name appearing. I've found just leaving the old station name means EliteOCR will update it correctly, but it's not intuitive.

While I was working on getting this running on Mac I remembered reading about pytesseract, but it looks like it's not the same thing as python-tesseract. Just wondering if you looked at both and selected python-tesseract for a particular reason, or if it was just the first thing you found.

I was interested in seeing if it would be feasible to port EliteOCR over to pytesseract because it seems to be better maintained overall, but I want to make sure I'm not wasting my time because it's lacking a feature that you need or anything like that.

Hi
Latest release seems to break the export directory. I have a sepaarate Export Directory from my screenshot directory, but when i "Export" its opening up a dialog in the screenshots directory not the export directory.
Thanks

Would it be possible to have this watch for new imaes, and as they appear, process them, updating a per system/station entry in a local DB?

In preparation for the Mac OS release due in a couple of months. Should be possible thanks to Python. Would be happy to fork and lend a hand if you like?

Hi,

i wanted to try EliteOCR under Ubuntu. Like mentioned in the description i installed all dependency (btw. BeautifulSoup4 is missing in the ReadMe) and i get this error in the errorlog.txt directly after executing the script.

2015-05-23 16:53:53,174 CRITICAL:
  File "EliteOCR.py", line 1215, in <module>
    sys.exit(main(sys.argv[1:]))
  File "EliteOCR.py", line 1208, in main
    window = EliteOCR(app)
  File "EliteOCR.py", line 84, in __init__
    self.settings = Settings(self)
  File "/home/dennis/Dokumente/Elite/EliteOCR/settings.py", line 24, in __init__
    self.values = self.loadSettings()
  File "/home/dennis/Dokumente/Elite/EliteOCR/settings.py", line 57, in loadSettings
    'horizontal_exp': self.reg.value('horizontal_exp', type=bool),

<type 'exceptions.TypeError'>: unable to convert a QVariant of type 0 to a QMetaType of type 1

Hi @seeebek, do you have a file on your system within <tesseract>/share/tessdata named big.traineddata? I am getting errors that it doesn't exist (which it doesn't, I only have english and osd). Just wondering if it was created by windows or I need to obtain it from somewhere?

Cheers

Complete date plus hours, minutes and seconds:
YYYY-MM-DDThh:mm:ssTZD (eg 1997-07-16T19:20:30+01:00)

Complete date plus hours, minutes, seconds and a decimal fraction of a
second
YYYY-MM-DDThh:mm:ss.sTZD (eg 1997-07-16T19:20:30.45+01:00)

Without the time-zone info, when I receive an update over EDDN I don't know what the time really is.
(I have dates in the future in QAA now!)

@DullingWine: it doesn't seem possibly to change the station name if its misread. e.g. Miller Depot was read as Miller Oepot, but when I change it in the station box, it doesn't then carry that change forward to the station field in the commodity lines.

Awesome work!

I wonder if it's possible to use EliteOCR via a bash. Is there a simple CLI call? Something like ...

$ ocr.py Screenshot_0002.bmp

which just does the OCR and outputs the pricings in some kind of data structure like JSON, XML or something like that?

Would be really nice.

Happens mostly when market is scrolled all the way to the bottom, and some of the first goods are also present in the previous screenshot.

errorlog.txt:

2015-02-08 16:34:50,950 CRITICAL:
n  File "C:\Users\SEBAST~1\Desktop\RFACTO~2\build\EliteOCR\out00-PYZ.pyz\export", line 194, in eddnExport

<type 'exceptions.TypeError'>: arguments did not match any overloaded call:
  QMessageBox.warning(QWidget, QString, QString, QMessageBox.StandardButtons buttons=QMessageBox.Ok, QMessageBox.StandardButton defaultButton=QMessageBox.NoButton): argument 1 has unexpected type 'instance'
  QMessageBox.warning(QWidget, QString, QString, int, int, int button2=0): argument 1 has unexpected type 'instance'
  QMessageBox.warning(QWidget, QString, QString, QString, QString button1Text=QString(), QString button2Text=QString(), int defaultButtonNumber=0, int escapeButtonNumber=-1): argument 1 has unexpected type 'instance'

Example of overlap
SS1: https://a.pomf.se/sfnpxw.png
SS2: https://a.pomf.se/hauuxx.png

Exporting normally and to BPC does not seem to produce any errors.

To overcome this I have been taking a screenshot and exporting everything except the final part of the market, and then exporting just that final part individually.

another suggestion to ease the use of EliteOCR:
In the learning wizard screen, make the text output larger and put it above the black/white images.

Reasoning behind the suggestion:
What we do is: we look at the original/inverted image and then we check the detected text which is the text output. BUT there are large pictures of single characters in between which draw our attention away from the scanning result towards the image that was used for scanning. This image is only useful as a secondary reference. It is not that useful for checking if the scanning result is correct. The small characters which are also spaced widely apart make it really hard to read the scanning result.

The dev version that has the option to auto-delete processed files leaves the last file in the list and not deleted. It looks like maybe the delete happens during the beginning of the processing of the next file. At the end because there isn't another file afterwards it just disables the buttons and leaves things in the list.

I got an error after running eliteOCRcmd.exe -i ./test/screenshot_0000.bmp -o test.csv -l eng
The program runs fine with the GUI but I want to include the command line tool in a file watcher and automatically upload the results.

Error:

Please make sure the TESSDATA_PREFIX environment variable is set to the parent directory of your "tessdata" directory.
Failed loading language 'big'
Tesseract couldn't load any languages!
AdaptedTemplates != NULL:Error:Assert failed:in file adaptmatch.cpp, line 174

Everyone agrees that EliteOCR works brilliantly for the Traders in Elite. But how about some love for the Explorers, eh? You ever met an explorer that doesn't like to record their Cartographic Data?
But doing it manually is a slow and tedious process. Could EliteOCR help?

Please look at the attached screenshot for an example of Cartographic Data that could be OCR'd and exported to other apps like the great add-on tool for Elite: EDDiscovery. I've highlighted in purple the bits of data that would be interesting.

As with the OCR program currently operates with screenshots of the market in a set folder, likewise we could configure where the Cartographic screenshots are held in a different folder so as not to mess up the original functionality.

I'm not a programmer so no idea how difficult to implement this would be, sorry.

I am currently having to update a lot of the figures manually and am struggling to read the figures from the preview panel. It might be my eyesight or screenshot resolution - I will play with that separately (resolution not eyesight :) ) but as I also run the OCR software on my laptop (I have access to the screenshot folder over the network) I often am using a smaller screen. Therefore, if possible it would be nice to add a zoom to the panel and allow panning with the mouse (i.e. click and drag - imo this is best solution) or scroll-bars.

Again, not a high priority problem (just an enhancement) and if I get a chance I will have a look at implementing this but python is not my most familiar language

Due to the new bars representing stock levels being introduced in the new versions this week, that have replaced the Low/Med/High text, the stock values are always returned as Zero.

Also they're many new commodities that have been added in the new versions.

I wanted to do it hard way and run it from the source, so I went through the list of dependencies and on top of what is already mentioned in readme I had to install:

pip install pytz
pip install tzlocal
pip install beautifulsoup4
pip install scikit-learn (did not work through pip, installed from pre-built package)
pip install scipy (did not work through pip either)
pip install requests
pip install wget

I am on Windows 7 64-bit, running Python 2.7.9 32-bit, I have VC environment installed, so technically I should be able to build any C module too. I believe I installed superpack of scipy, which probably pulls in more stuff than necessary (possibly numpy as well), but it seemed to be already too much of mess trying to install scipy "pure".

I departed from official Python install from python.org. I guess maybe some of the packages listed above are already present in some other Python distros, but for what concerns the vanilla install, they are not.

How about you just add all the fields the user manually corrects to the training examples?

The reason behing this suggestions is the following:
The most convenient way to use the program is "add all, ocr all, automatically delete files upon completion" BUT if you use the program like this you do not have the file which had ocr errors in it to learn from it.
ALSO: We check the ocr results of the file to make sure no wrong data is uploaded and then we have to check all of it again if there were some errors in it.
Both these circumstances create much more work and take away the fun of playing the game.

Please just let us add single words/numbers to the training pool and automatically add the characters we corrected. in the table or in the preview field.

I just noticed that 0.5.4.1 has been announced on the forums but I'm not seeing any new code here. Forget to push? 😃

I have found when using EliteOCR there appears to be an issue when pressing the "Add and next" button after correcting an incorrect figure/adding a missing one

This image shows the data missing from the the OCR data:

The next image shows that I have entered the missing data (the demand quantity) and pressed the "Add and next" button, the data has been added to the table but the blue rectangle is still on the same commodity (FISH) - (and yes I missed that I needed to check the sell price :) )

The final image shows the second press where the blue rectangle has now moved onto the next commodity (FOOD CARTRIDGES)

This happens whenever I need to modify or input new figures and the value does not appear in the drop down list (if it does the problem is not present and the program moves onto the next line)

Obviously not a high priority bug, but just thought I would make you aware - If I get a chance (and can understand the code) I will try and see if I can resolve it myself

Cheers

Codersparks

My screens aren't very high res so F10 screenies don't OCR very well. Hence I try alt-f10 screen shots out generating 83 meg files, but when I attempt to OCR them I get (every time):

Traceback (most recent call last):
File "", line 175, in addFiles
File "", line 228, in selectFile
File "C:\Users\SEBAST1\Desktop\elite\RFACTO1\build\EliteOCR\out00-PYZ.pyz\cu
stomqlistwidgetitem", line 33, in loadPreviewImage
File "C:\Users\SEBAST1\Desktop\elite\RFACTO1\build\EliteOCR\out00-PYZ.pyz\cu
stomqlistwidgetitem", line 54, in addPreviewImage
File "C:\Users\SEBAST1\Desktop\elite\RFACTO1\build\EliteOCR\out00-PYZ.pyz\oc
rmethods", line 18, in init
File "C:\Users\SEBAST1\Desktop\elite\RFACTO1\build\EliteOCR\out00-PYZ.pyz\oc
rmethods", line 42, in findAreas
MemoryError

Any suggestions?