The project started with building a cloud camera using raspberry pi. At the end of it, we had a fully functional cloud camera on the raspberry pi platform. Sold and installed several pieces; many were installed in the neighborhood. Following observations were made:
Given the learning, we decided to do another round of improvements to cater to the Indian customers. Creating a 2Module doorbell camera that can replace the existing electrical doorbell seemed like a good entry point, as there is no other product catering to this segment. The smart cloud doorbell is the result of this effort.
The doorbell camera conforms to the expected functionality. In cloud mode it is constantly checking for motion and uploads snapshots when motion is detected. The bell button-press on the smart doorbell camera sends a Push Notification to the mobile App, either Android or iOS; usually, it hits the phone within 2 seconds. Clicking the push notification will open the live video and motion history view in the respective mobile App. Motion history lets you know who pressed the bell even if they have moved away. The live view gives you the current video feed from the bell with very low latency.
The bell can send notifications to both Android and iOS phones. The bell starts recording history on the cloud as soon as it is configured and running. The bell video feed and history can be accessed from any browser or mobile apps.
|Doorbell press alert on iWatch||2M doorbell with camera||2M PCB stack|
The bell is low-cost and comes with high-quality software and hardware. It is built keeping Indian customers in mind. It comes with night-vision and a relay to drive one external doorbell. it will fit in as any other 2module device like a smart switch or a fan regulator. With wide angle camera it will cover the full room from where it is installed. It can also ring other Ibeyonde ringers residing on the network using UDP broadcast of bell ring on the local network.
The cloud cameras are fully accessible from all the popular browsers from desktops and laptops. In addition there are Android and iOS Apps. Firebase is configured to send push notifications to the mobile Apps. The initial configuration is done via bluetooth; on reset it reverts to the same.
At the backend, resides a pipeline of AI/ML technologies. These are object detection, including faces and dangerous objects, object counting and license plate number recognition. Face recognition can then be enabled and configured using a process of naming detected faces. These were inherited from the cloud camera. The camera API lets you request the visual data from the bell camera on demand. This enables cloud API that lets you query camera for detected objects, count, and scenarios in real time. This drastically reduces the networking and storage costs.
The cloud functionality can be enabled or disabled. If disabled, the bell is only accessible from the home network, and there is no recorded history.
Manually initiated wireless(FOTA) upgrades are available.
IR cameras used in the door bell are tuned for wider range of frequencies in order to provide vision in low or no light conditions.
|wide-angle doorbell camera||night view||lit room view|
|regular camera||night view||lit room view|
The motion pictures captured when QR-code from the phone is flashed in front of the camera
QR-code interpreted, bell rings
Similar doorbells with less features are priced upwards of 4k. We should be able to get it done in about MAX 1k that will give us enough room to price it competitively.