This program was developed for a smart basket that runs on a Raspberry Pi. It is built to participate in the GPM's 11th National College AI Intelligent Automation Equipment Creation Award competition. Among 67 participants, we rank in the top 21. These are the details for the final announcement for all the projects that have been accepted (our team is listed under project number 12).
The following is a poster that we provided for this project.
The program was developed under the Raspberry Pi environment.
Install the dependencies
python -m pip install --user -r .\requirements\requirements.txt
You will also need to install another dependency in the .\requirements/installing_pyaudio.txt file.
Train the model
python .\codes\train\train.py
In addition to training the model, melfrequency features will also be extracted into the pickle and saved for later use. In order to avoid re-extraction of features when retraining the model, you may wish to comment off line 54 of the train.py file. The process of extracting the features will take time.
The folder model contains a saved model. The following command can be executed to test the model with the available testing data.
Test the model
python .\codes\test\test.py
Data collection (adjust the class that you want to collect inside soundcollection.py)
python ./soundcollection.py