Cameras watch key areas
Place cameras in rooms, hallways, stair areas, or other high-risk zones where a fall may happen.
How it works
SafetySense FallShield uses affordable cameras and local AI processing on a Windows PC or laptop. For 1–2 cameras, testing can work on an ordinary computer without a dedicated GPU; a GPU is recommended as camera count grows. Video stays on site. The cloud is used for account, subscription, device status, and alert delivery — not for live video processing.
Place cameras in rooms, hallways, stair areas, or other high-risk zones where a fall may happen.
The FallShield computer analyzes motion and posture locally. Private video does not need to leave the home.
The system focuses on meaningful fall-like events instead of sending raw camera notifications.
With a trial or paid subscription, SafetySense can forward alerts by configured channels such as email or SMS.
For demo testing and simple setups with about 1–2 cameras, FallShield can be tested on an ordinary Windows laptop or PC, including integrated and USB cameras. In our CPU-only test on an Intel Core i5-8250U @ 1.60GHz, reaction time was acceptable for 1–2 cameras.
When you use more than two cameras, an NVIDIA GPU is recommended for smoother continuous monitoring. CPU-only processing can still work, but recognition may be delayed by tens of seconds and some CPUs may run hot depending on the model.
In testing mode, wait for the recognition result for up to a few minutes before deciding that a test failed. In real-world use, a short delay is usually still acceptable because the system continues watching and processing locally.
Demo mode is for local testing. It runs on the installed computer and shows on-screen notifications without requiring an account. It is also useful for checking whether your existing CPU-only computer is fast enough for your camera count.
A subscription connects the installed device to a SafetySense account so alerts and account-managed settings can be enabled.
Wearables can help, but they only work when the person is wearing and charging them, and some falls may not create the impact pattern a watch expects. SafetySense uses room-based visual AI, so it can monitor high-risk areas continuously without asking the person to remember a device.
The video above compares an earlier FallShield demo with a wearable-style approach. It is best understood as a concept demonstration, not as a replacement for medical advice or emergency services.
Download the Windows installer and run local demo mode first to confirm camera coverage.
Detection quality depends on camera position, lighting, and avoiding blind zones.
Create an account and start a trial or paid subscription when you are ready to forward alerts.