Falls are the one of the leading causes of injury for the elderly. Every 11 seconds, an older adult is treated in the emergency room for a fall, and every 19 minutes, an older adult dies from a fall.* With such shocking statistics, how could we not help the elderly?
Looking around, we can see there are already products like Life Alert that help with falls. However, most of these products are wearable devices, which is problematic because the elderly may find such products inconvenient to wear. Sometimes, they may entirely forget to wear the devices. Thus, we decided to create a method to automatically detect falls and quickly bring assistance to the elderly.
What it Does
SafeLine uses a Raspberry Pi camera to continuously scan for falls. Once a fall is detected, SafeLine immediately notifies connected users by sending them an email and pushing a notification to their iPhones through an iOS app.
How It Works
SafeLine consists of a number of components. On a Raspberry Pi, we have connected a camera and used a Python script with OpenCV to continuously check for falls. Once the algorithm detects someone falling, it uses the SendGrid API to email a message and an image of the fall to a list of connected users. It also uploads a timestamp and the image to Firebase. Firebase also stores the list of emails for the Python script. Meanwhile, on each user's iPhones, an iOS app, built with Swift and XCode, detects that Firebase has changed. It then creates the notification and shows it on the user's phone.
Finally, using React.js, we set up a website to describe our project and motives on Github pages. The website is also registered at safeline.tech.
Along the way, we encountered numerous challenges. Perhaps the biggest challenge we had was developing the iOS app. None of us has experience with Swift or XCode, so we struggled to connect the app to Firebase and make it send push notifications. Another challenge involved improving the sampling rate on the Raspberry Pi's camera. Since processing images and sending the fall data requires a lot of computational power, the camera recording was beginning to lag a lot. Hence, we split the Python script into multiple threads to prevent separate tasks from interfering with each other. Finally, we found it challenging to design an appropriate logo and user interface that would represent our project. To solve this problem, we sought the help of both each other and the people around us. Overall, we are happy that we were able to overcome these challenges and connect our entire project.
What's next for SafeLine?
Here are some of the improvements we could make for SafeLine in the future:
- Multiple cameras - we had only one camera for this project, and we would certainly like to integrate several, like a real home would have
- iOS app "friends" - we could add features such as a "friends list" where people can request to watch over someone else, or to have someone else watch over them
- Better accuracy - the detection algorithm has much room for improvement, such as in minimizing the rate of false positives
- Location detection - people who are near the person who has fallen can be notified, even if they do not specifically monitor that person
- Sound detection - on top of image detection, we could learn to recognize sounds of falling or cries for help