Guard Dog

Inspiration

Keep life's dangers in the Dog House

What it does

Runs in background, monitoring frames of accelerometer data. Runs these frames through a pretrained machine learning algorithm. If it determines that the acceleration in the past few seconds is cause for concern, it will prompt you with a dialog box asking if you are okay. If you fail to respond, it will reach out to your "Guard Dogs" – people who you have registered with the app as those you would like to keep informed of your safety. It can also place a phone call to one emergency contact.

How we built it

Love

Challenges we ran into

Pebble proved to be too difficult to integrate into the project

Accomplishments that we're proud of

Still integrated on Android Wear, accomplished all original goals

What we learned

^{_{(Never trust the Pebble team to build a sensible API)}} redis is a really cool database stand-in; it's got a really simple API and great feature sets for storing key-value pairs. We also gained insight that scrum is a great way to make hackathons less stressful and much more organized.

What's next for Guard Dog

• We'd like to make the project more robust, user-interface wise
• We'd also like to make it more battery-efficient
• Explore new machine learning algorithms and tweaks to make our data classification more accurate

Built With

• android
• bootstrap
• python
• redis
• scikit-learn