Distributed Smart Camera System
Faculty Mentor
Arindam Das
Document Type
Poster
Start Date
10-5-2023 9:00 AM
End Date
10-5-2023 10:45 AM
Location
PUB NCR
Department
Engineering
Abstract
In the area of home security, inexpensive camera systems are becoming increasingly capable with the use of machine learning. When paired with server-side machine learning algorithms, even entry-level smart cameras can be used for object detection and image classification. A typical consumer can benefit from these systems with a WiFi connection and AC power source; however, many use cases lack one or both of these criteria. Our objective is to create an affordable security system for environments without reliable access to Internet access or electrical power. To make our system viable, it must meet several criteria. It must be operable without an external power source, it must have a wireless range of at least 1 mile, and it must not be prohibitively expensive to manufacture. To meet these criteria, we are attempting to implement a machine learning algorithm which categorizes objects and events prior to sending image or video data over the wireless network. If each remote camera in the distributed system can determine which events can be safely ignored, i.e., a tumbleweed, the average power consumption of the device drops. Lower power consumption results in a smaller battery and solar panel needed to power the device, further decreasing its cost. In order to achieve these goals, we are writing software for a low-power microcontroller and designing a circuit board to create a minimum viable product of a long-range smart camera with no external power source.
Recommended Citation
Norman, Mark and Liebert, Justin, "Distributed Smart Camera System" (2023). 2023 Symposium. 14.
https://dc.ewu.edu/srcw_2023/res_2023/p1_2023/14
Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-No Derivative Works 4.0 International License.
Distributed Smart Camera System
PUB NCR
In the area of home security, inexpensive camera systems are becoming increasingly capable with the use of machine learning. When paired with server-side machine learning algorithms, even entry-level smart cameras can be used for object detection and image classification. A typical consumer can benefit from these systems with a WiFi connection and AC power source; however, many use cases lack one or both of these criteria. Our objective is to create an affordable security system for environments without reliable access to Internet access or electrical power. To make our system viable, it must meet several criteria. It must be operable without an external power source, it must have a wireless range of at least 1 mile, and it must not be prohibitively expensive to manufacture. To meet these criteria, we are attempting to implement a machine learning algorithm which categorizes objects and events prior to sending image or video data over the wireless network. If each remote camera in the distributed system can determine which events can be safely ignored, i.e., a tumbleweed, the average power consumption of the device drops. Lower power consumption results in a smaller battery and solar panel needed to power the device, further decreasing its cost. In order to achieve these goals, we are writing software for a low-power microcontroller and designing a circuit board to create a minimum viable product of a long-range smart camera with no external power source.
