mitchellss/cvgui

cvgui is a Python library for creating body-interactive GUIs using computer vision.

cvgui's Issues

Change from xcoord/ycoord to pos=(x,y)

Using a tuple to define the position of a component seems like a better way of doing things instead of having separate values for each axis.

Move from setup.py to setup.cfg

Creating packages with setup.py is going to be removed for future versions of pip so this needs to be changed.

Add scene switching

Currently, the program can only display the current scene. This hardwired option needs to be removed and replaced to allow for dynamic scene switching.

Feedback device stubs

Need to add stubs (and maybe a new example executable) for providing basic feedback to a user

Make working example more appealing

The example code needs to have the skeleton centered and have the buttons bounce around the screen when hit. Right now the example is a bit lackluster.

Add windows compatability

Currently, the package only works on Linux due to the differences in Python multiprocessing on Windows and Linux.

Add radius attribute to button

Allow custom coloring of components

Custom skeleton model positioning and scaling

Currently, the skeleton model in the UI cannot be scaled or moved. The scale and position of the skeleton should be provided by arguments when creating the skeleton.

Add intensity level as blazepose parameter

Currently, the blazepose intensity is hardcoded as zero. Most systems can do better than this so the intensity should be a parameter of the class.

Add ability to record pose data

There needs to be a way to record pose data and save it to a specified format. Starting with being able to save to a CSV format is probably a good start.

Add per-frame callback to component protocol

All components that need to do something every frame (move to position, check for collision, etc) should have these methods in a predefined function that fulfills the component protocol instead of having the activity class do separate checks for each component type.

Move from pdoc3 to pdoc

Add scene "enter" and "exit" functions

In the same way a scene has a "per-frame" callback I think it would be useful to have "enter" and "exit" callback functions. These would be useful for starting/stopping loggers, re-positioning components, and cleaning up user-instantiated objects between scenes.

Encapsulate activation distance in button component

Currently, the activation distance is passed to the is_clicked method as a parameter. Instead, the button should just know its own activation distance and position and determine if it is clicked based on a passed pair of coordinates.

Implement hand bubble component

To improve the user experience of UIs created with the tool, a component that follows the user's hands around needs to be implemented. Ideally, these "bubbles" are then made to be the same color as the buttons they can click.

Lagging issue

Using Windows built-in camera there is a bit of a lag delay when doing the activities.

Implement text label component

There needs to be a way of displaying text on the screen to the user.

Replace getters/setters with attributes/properties