Image recognition to automate drive testing

Question

Are there any tools that use Image Recognition (search, comparison, image verification) as a base for automation and testing of GUI software. I know ranorex supports it. Are there any more efficient tools? Are there any problems using Image Recognition to automate test testing?

Answer 1

Well, first of all, I understand the previous answers: testing applications using pattern recognition is not the best way to test GUIs. But at the same time, I don’t understand why you are not answering the first question. He asks for tools that work this way, I think he is smart enough to understand where he goes.

Ok, now the main question, my choice will include:

• Sikuli , a MIT project under a MIT license similar to GNU. It uses Python over Jython. Free.
• TestPlant eggPlant is a tool that works through a VNC server, so you can test applications on any VNC compatible platform (including smartphones). It has some interesting features, such as OCR, test schedule, and so on. It uses SenseTalk. Not for free, you can request a trial version.
• Routine Bot , I have never used it, but it seems very useful.

Answer 2

I will also prevent the use of Image Recognition with SendKeys and Click at Coordinates or (Button Images) to test the user interface. I recently used UI Automation to successfully automate testing a WPF application. By placing small breadcrumbs (Automation.AutomationID = "OkButton") throughout our XAML application, I was able to write several C # tests that perform different aspects of the application. Even without breading, UI Automation can still run the application, but trying to identify controls in the user interface is a bit more complicated.

A decent article on Code Project is available as a starting point.

http://www.codeproject.com/KB/WPF/UIAutomation.aspx

You will also need UI Spy, a free tool from Microsoft to help you find controls and manually use controls through UI Automation as a guide for writing scripts. The tool is buried in the Windows Vista SDK after searching for an installation for UISpy.exe. The UI Spy tool can still work on a computer running Windows XP by simply copying the EXE to the target machine.

Answer 3

Let me suggest a different solution.
This is not a complete UI automation infrastructure, but rather a specific tool for image verification only.
This will allow you to ignore the unstable part of your images (random data, etc.)
It will integrate with any other user interface testing platforms you choose: Selenium, Sikuli, etc.

http://visualci.com

Answer 4

Thank you for your comment! please check RoutineBot software - an interface based on clicking on specific image templates and see for yourself how this idea is implemented into an automation tool!

Answer 5

I agree with RodKnee and Tom E. Using image recognition to test the user interface is nothing less than a painful and wasteful time. If your application is built correctly, you can push the user interface logic to a different level of your application, which will be more easily tested.

An MVP template has been created for this purpose. Each operation that can be performed through the user interface is presented somewhere in the view. This way you can completely remove the user interface and unit test your application using the remaining architecture (MVC).

The thought of image recognition to achieve this kind of thing makes me tremble.

Answer 6

Consider AutoItScript for managing Windows-based GUIs in test cases - And scramble user interfaces. Consider the open source optical character recognition tesseract. Also OpenCV for machine vision.

Free AutoItScript works at the API level, because you can read the states of various sections of Widgets and Windows, send actions to these components of the user interface, wait for state changes, etc. It is possible to create high-strength automation code that provides focus on Windows and resolution independence.

Answer 7

An old question, but perhaps this answer may be useful to someone. I am currently using two products,

Testing anywhere with Anywhere Automation ( http://www.automationanywhere.com/Testing/ )

and Quick Test Professional, HP ( http://www8.hp.com/us/en/software-solutions/software.html?compURI=1172957#.UhJBwpLW5-k )

Both of them do the job well enough, and both support the use of pattern recognition. I'm not quite sure that pattern recognition in itself is bad. As in all other cases, you must adapt your approach to your specific needs and use the right tool for the job.

Answer 8

Just thought I'd add another entry to this thread. Maybe everything has changed, but I'm not sure, but when I last saw the demo, this product offered Sikuli IDE / interface / features as a commercial product and supported real devices outside the simulator. I don’t know if the tool for detecting objects by identifiers outside the images has improved now or not.

SeeTest from http://experitest.com