update(docs): document confidence level

README.rst

@@ -6,6 +6,11 @@ This Robot Framework library provides the facilities to automate GUIs based on
 image recognition similar to Sikuli. This library wraps pyautogui_ to achieve
 this.
 
+For non pixel perfect matches, there is a feature called `confidence level`
+that comes with a dependency OpenCV (python package: `opencv-python`).
+This functionality is optional - you are not required to
+install `opencv-python` package if you do not use confidence level.
+
 Keyword documentation
 ---------------------
 
@@ -85,7 +90,6 @@ You additionally need to install these for pyautogui_:
 ::
 
     $ sudo apt-get install python-dev python-xlib
-    $ sudo pip install pillow
 
 
 You might also need, depending on your Python distribution, to install:
@@ -100,24 +104,46 @@ virtual environment for pyautogui_:
 - `Fetch the source distribution`_
 - Install with:
 
-  ::
+::
 
-        $ pip install python-xlib-<latest version>.tar.gz
+    $ pip install python-xlib-<latest version>.tar.gz
 
-Running tests
--------------
+Running unit tests
+------------------
 
 ::
 
     $ python tests/utest/run_tests.py [verbosity=2]
 
-and
+
+Running acceptance tests
+------------------------
+
+Additionally to unit test dependencies, you also need OpenCV, Eel and Chrome/Chromium browser.
+OpenCV is used because this tests are testing also confidence level.
+Browser is used by Eel for cross-platform GUI demo application.
+
+::
+    $ pip install opencv-python eel
+
+
+To run tests, run this command:
 
 ::
 
     $ python tests/atest/run_tests.py
 
 
+Updating Docs
+-------------
+
+To regenerate documentation (`doc/ImageHorizonLibrary.html`), use this command:
+
+::
+
+    $ python -m robot.libdoc -P ./src ImageHorizonLibrary doc/ImageHorizonLibrary.html
+
+
 .. _Python 3.x: http://python.org
 .. _pip: https://pypi.python.org/pypi/pip
 .. _pyautogui: https://github.com/asweigart/pyautogui

src/ImageHorizonLibrary/__init__.py

@@ -44,12 +44,31 @@ class ImageHorizonLibrary(_Keyboard,
     facilities to recognize images on screen. It can also take screenshots in
     case of failure or otherwise.
 
+
     This library is built on top of
     [https://pyautogui.readthedocs.org|pyautogui].
 
+    == Confidence Level ==
+    By default, image recognition searches images with pixel-perfect matching.
+    This is in many scenarios too precise, as changing desktop background,
+    transpareny in the reference images, slightly changing resolutions, and
+    myriad of factors might throw the algorithm off. In these cases, it is
+    advised to adjust the precision manually.
+
+    This ability to adjust can be enabled by installing
+    [https://pypi.org/project/opencv-python|opencv-python] Python package
+    separately:
+
+    | $ pip install opencv-python
+
+    After installation, the library will use OpenCV, which enables setting the
+    precision during `library importing` and during the test case  with keyword
+    `Set Confidence`.
+
+
     = Reference image names =
     ``reference_image`` parameter can be either a single file, or a folder.
-    If ``reference_image`` is a folder, image recognition is tried separately 
+    If ``reference_image`` is a folder, image recognition is tried separately
     for each image in that folder, in alphabetical order until a match is found.
 
     For ease of use, reference image names are automatically normalized
@@ -71,9 +90,8 @@ class ImageHorizonLibrary(_Keyboard,
     data:
 
     | `Import Library` | ImageHorizonLibrary                   | reference_folder=images |                                                            |
-    | `Click Image`    | popup Window title                    |                         | # Path is images/popup_window_title.png                    | 
-    | `Click Image`    | button Login Without User Credentials |                         | # Path is images/button_login_without_user_credentials.png | 
-
+    | `Click Image`    | popup Window title                    |                         | # Path is images/popup_window_title.png                    |
+    | `Click Image`    | button Login Without User Credentials |                         | # Path is images/button_login_without_user_credentials.png |
 
     = Performance =
 
@@ -88,7 +106,7 @@ class ImageHorizonLibrary(_Keyboard,
     In the above example, same image is located twice. Below is an example how
     we can leverage the returned location:
 
-    | ${location}=           | `Wait For`  | label Name | 
+    | ${location}=           | `Wait For`  | label Name |
     | `Click To The Left Of` | ${location} | 200        |
     '''
 
@@ -242,10 +260,12 @@ def set_screenshot_folder(self, screenshot_folder_path):
         self.screenshot_folder = screenshot_folder_path
 
     def set_confidence(self, new_confidence):
-        '''Sets the confidence level for finding images
-        Applicable if opencv (python-opencv) is installed.
-        Allows for setting the value to None if you don't want
-        to use it or a value between 0 and 1 inclusive.
+        '''Sets the accuracy when finding images.
+
+        ``new_confidence`` is a decimal number between 0 and 1 inclusive.
+
+        See `Confidence level` about additional dependencies that needs to be
+        installed before this keyword has any effect.
         '''
         if new_confidence is not None:
             try: