Hand Music Changer is a tool for changing the music with your hands trough your camera. It uses a CNN to detect your hand and change the music.

• 🎶 Hand Music Changer ✋👈✊
  • How is it made?
    • Dataset
  • How to use it?
    • Hide/Show GUI
  • Hand signs
  • Configuration
    • Tolerances
    • Selected camera
    • Time gap
  • Build an executable

The CNN trained for the app model is based on MobileNetv2 from Keras. This approach allows us to transfer its learning for our purposes. You can check out the notebook for more information and technical details.

The model uses three hand gestures from the dataset and another class of random images for the negative samples. The gestures used are shown below:

You must use Python 3 for the entire project.

There are a lot of requirements to use it due to there is no executable file created yet. Those requierements can be installed with pip as follows:

python -m pip install imutils keyboard numpy opencv-python pillow pyautogui tensorflow

Or just install the requirements.txt file.

python -m pip install -r requirements.txt

Then, execute the following command to start the app:

python ./hmc.py

Notice that the app depends on several files located at ./hmc/. So, you must make sure that those files are in the same directory as the app.

There is a button to hide the camera feed. You must use Ctrl+Alt+Shift+D to show/hide the GUI. So it will not distract you.

Play/Pause ✋	Next song 👈	Previous song ✊

It is worth say that it does not matter if you flip the gesture, it will work because of the data augmentation during training.

Certainly there are a variety of environments to use the app. That is why there are some configuration options.

You can adjust the tolerances and the time gap to make the app more accurate. For instance, if play/pause does not work properly, you can try decreasing the tolerance for this hand sign, so it will be triggered with less confidence score. But, if the app suddenly changes or pauses the song you should increase the required confidence score to trigger the action.

You are able to change the camera that is used to capture the video. The default camera is the first one. With this option you can use other camera for other purposes and other one dedicated to the app.

Finally, you can change the time between actions. This is useful whether you want to trigger the action too often or not.

It is a official release coming soon. Meanwhile you can create an executable using pyinstaller. The specs are already defined, just execute the following command:

pyinstaller HandMusicChanger.spec hmc.py