OpenCV and MicroMacro

[BlubBlab]

Check list:
-licence
-removed logs
-add manifest
-removed plugins directory
-updated source code in src

64-bit version: http://www.mediafire.com/download/lv2cu ... 17_x64.zip
32-bit version: http://www.mediafire.com/download/xdq7u ... 17_x86.zip
changes:
-fix windows.pixelSearch (merge with admins version)

-added cv.getFilterPixel(hwnd, int x, int y,)
either get the Pixel from hwnd windows and set the filter or if a filter is defined it will get the pixel out of it
returns red,green,blue,alpha

-added cv.setFilter(hwnd)
made a screenshot of the hwnd window and load it in the filter directly.

-added (experimental) cv.findCorners(hwd[, int lenght][, max_gap )
seeks an object with the numbers of corners given with length
returns true or false. This is an independent project from me and needed to ad a new dll.

I will add a pixel search in the cv module later

[BlubBlab]

Check list:
-licence
-removed logs
-add manifest
-removed plugins directory
-updated source code in src

(32-bit) http://www.mediafire.com/download/i52l1 ... 19_x86.zip
(64-bit) http://www.mediafire.com/download/7lfwb ... 19_x64.zip

changes:
-fix typo with cv.getFilterPixel()

-added cv.hasFilterBuffer()
return if a filterbuffer is active or not(bool)

-added /cv.getFilterBufferDimension()
This function return the dimension of the filterbuffer
Returns int cols(x),rows(y) dimension or nil

-added cv.pixelFilterSearch(number hwnd, number r, number g, number b,
number x1, number y1, number x2, number y2,
number accuracy)

Returns (on success): number x
number y or nil

Note: pixelFilterSearch can be used with a predefined filterbuffer or with hwnd. It will not a new filterbuffer created and the old one will not be deleted.

Other updates: added Codebreaker as repro, forged with MM version from admin on git, updated my version of rombot, updated my meta-botframwork

TODO on MM2 with OpenCV :
-testing pixelFilterSearch
-testing corner search and update the back library to use CV with Houge and HougeP

[BlubBlab]

Check list:
-license
-removed logs
-add manifest
-removed plugins directory
-updated source code in src
-updated git repo

(32-bit)http://www.mediafire.com/download/3klhn ... 23_x86.zip
(64-bit) http://www.mediafire.com/download/1ckv7 ... 23_x64.zip

Changes:
- merge with Admins bug fix
- starting to implement regex functions

TODO on MM2 with OpenCV :
-testing pixelFilterSearch
-testing corner search and update the back library to use CV with Houge and HougeP
- check the 64-bit version of OpenAl(needed replace with an old version but it works)

Removed:
- All the old links

[BlubBlab]

Check list:
-license(update)
-removed logs
-add manifest
-removed plugins directory
-updated source code in src (later it still not final)
-updated git repo

(32-bit) http://www.mediafire.com/download/8jwpy ... 27_x86.zip
(64-bit) http://www.mediafire.com/download/f3a4g ... 27_x64.zip

Changes:
- merge with Admins bug fix for memory
- smerged with sqlite support

TODO on MM2 with OpenCV :
-testing pixelFilterSearch
-testing corner search and update the back library to use CV with Houge and HougeP
- waiting for what admin does for sylite3

[Administrator]

I was unable to download the 32-bit binary as MediaFire said it was a "dangerous file" and blocked me from accessing it. Can you please email a copy to me?

[BlubBlab]

Check list:
-licence
-removed logs
-add manifest


Download : Frontpage

Changes:
-(fix)The alpha channel seems to be invalid(in the screenshot conversion to the matrix) I don't know if this was done through the update or open CV 3.1 or this was always problem and it randomly crashed.
- (fix) It seems there is a race condition with the mutex class when shutting down in the unload function I simple deactivated the exception handling and let it run otherwise it would randomly crash while closing the app.

-update to Lua 5.3.3
-update to OpenAL 1.17.x( I need to fix some problem on their side first see: https://github.com/kcat/openal-soft/issues/45)
-update to OpenCV 3.1
-update Objectsan (to OpenCV 3.1 no code changed)
-update sqlight to 3.13

-The new code is inside the zips I will update git later.

[Administrator]

The shutdown issue should be fixed in one of my latest commits. Since then I've not had any issue with it, and in fact have been working on a neat little side project that I'll discuss more soon.

[Exempt]

Is there any examples of this in use? I'm curious what this is really used for in something like Micromacro?

[Exempt]

I can't really figure out what this is for, seems to be used for things like facial recognition. Would this be used to see things in a client window? If so how would you even start scanning a client window and what type of resource would be used to pick out the item or monster... ?

[Administrator]

Yes that's exactly right. It could be used for any sort of object recognition, though isn't the most efficient way to go about things.

[Exempt]

Is it possible you could give a very simple example of capturing a notepad window and watching for a certain letter?

[BlubBlab]

Well for letter I would use something different there is a way for Open CV to use an external OCR software but I haven't implemented this

But the way the object recognition works is first you need to make a trainings file for open cv like this :
https://www.youtube.com/watch?v=WEzm7L5zoZE
http://note.sonots.com/SciSoftware/haartraining.html
(tool is in the zip file)

Then in MM you call

Code: Select all

 
cv.objects(hwd, string xml_path[,int min_width, int min_height][,int max_width, int max_height][,bool debugwindow]))
 -- means something like
local NumberofFounds =   cv.objects(0,"training.xml")
  --ps: 0 is the whole desktop

Last step is read the position

Code: Select all

local tab = {cv.objects_next()};

---the function returns up to 10 tables if more you need to call again
local x = tab[1].x
local y = tab[1].y
local width = tab[1].width
local height = tab[1].heigth

[FrankWhids]

New Feb 1 OpenCV support code has been posted to github, is there any documentation on how to use it?