Image Recognition AI

The Eye support image recognition of small screenshots. The image is first analyzed and broken down into many smaller areas of interest. The image to the left is the actual image and the image to the right contains the image analysis.

The Eye finds the best match closest to the current location.

The red target area needs to be confirmed by the blue areas in the image.A target is confirmed if at least ConfirmationLevel percent of all areas are found. The areas needs to be located relative to the target area to be valid.

Vector Based Image Recognition

EyeAutomate supports both pixel and vector based image recognition.

Change to vector based image recognition using the Recognition command like in the example below:

Recognition Vectors

Vector based recognition is a lot less sensitive to zoom and rendering details. For example, a script created in Chrome with 100% zoom will most likely still work in the Firefox browser using 150% zoom.

Make sure to include some vertical or horizontal lines in the images when using vector based image recognition. The example below is a good example:

Either is used by default and requires either a vector or a pixel match. Both requires that both vectors and pixels match the target.

Multi-Threaded Image Detection

The Eye uses one thread per available CPU core when finding an image. Reduce the number of threads used with the CpuUsage command.

Multi-Screen Support

EyeAutomate and EyeStudio can both capture image and run scripts on any monitor connected to the machine. Note that using multiple screens reduces the run speed.

Manual Tests

Some steps in the test case might be difficult to automate. In EyeAutomate you may create manual tests by adding an empty step. EyeAutomate will show the Manual Test dialog containing the manual test instruction. The user selects the Pass, Fail or Fail All button after trying to perform the instruction. The Fail All button will also fail any other manual test dialogs that might appear. A user may also enter a comment in the Result field.

Begin “Verify that…”

Manual Recovery

EyeAutomate may ask the user to complete a failed command or Step manually instead of failing the test when manual recovery is enabled. Semi-automated tests are less fragile than fully automated tests. The user may, for example, be instructed to manually click on a button that EyeAutomate failed to find.

See the ManualRecovery and Begin commands for more information.

Automatic Recovery

To make a script even more reliable it is possible to insert one or many Catch commands between the Begin and End commands. The commands between the Catch command and the next Catch or End command will be executed automatically before recovering manually if the script fails before reaching the Catch command. You may insert any number of Catch commands and EyeAutomate will try them one-by-one until one of them is executed without problems.

See the Catch command for more information.

Encrypted Text

Insert encrypted text using the Insert Encrypted Text option in the Script menu. Encrypted texts are automatically decrypted by the Type and Paste commands. An encrypted text must be provided as a separate parameter to the command to be decrypted, for example:

Type “007a280819d700”