BBall Smart Net

Inspiration

I wanted to inspire google to allow non-speech audio files for machine learning model training. The fact that speech is required for audio analysis is highly limiting. Tracking non-speech audio could help monitor anything from industrial machinery to recreational basketball performance. The models could be implemented similar to the image recognition models, or a more novel approach would be to allow users to create a language/ array of sounds eligible to be converted to specified text as they occur.

What it does

To accommodate the speech requirement, I rigged up a raspberry pi with an ultrasonic sensor to track ball proximity to the net, and emit an mp3 of someone speaking to be picked up by google's synchronous speech to text api. The frequency of the speech entries would be compared to the number of minutes played to give a made shot per minute statistic. This information is to be stored in the users mobile device to be compared to future sessions.

How I built it

I used a raspberry pi with an ultrasonic sensor to monitor ball proximity to the net. I used a fork of google's speech to text github repository to monitor frequency of the person speaking through the mp3.

Challenges I ran into

Finding an HDMI cord to connect the raspberry pi to a monitor. Lack of supply of the required wiring components to set up the raspberry pi.

Accomplishments that I'm proud of

I am proud that the innovation I am suggesting could produce large amounts of revenue

What I learned

I learned a lot because this was the first time I set up a raspberry pi with a peripheral sensor, and the first time I used python.

Built With

• google-cloud
• python
• raspberry-pi