SecureVis
Format
SOECS Senior Project Demonstration
Faculty Mentor Name
Shon Vick
Faculty Mentor Department
Computer Science
Additional Faculty Mentor Name
Osvaldo Jimenez
Additional Faculty Mentor Department
Computer Science
Abstract/Artist Statement
SecureVis is an open source smart computer vision driven security system that leverages an edge computing paradigm and deep learning computer vision algorithms to determine human activity across multiple spaces. Multi camera systems can be configured across various rooms / spaces, allowing for maximum coverage across a users property. Small camera clusters can be configured to communicate with a microcomputer (e.g. raspberry pi) which uses low computational cost OpenCV algorithms to determine if there is considerable movement within each cameras optical view. If such movement is detected data is passed over the network to a master system which utilizes a GPU accelerated Regional Convolutional Neural Network (e.g. YOLO) to determine if the subsequent activity is human. This human activity then triggers the affected cameras to stream security footage from their respective edge systems to the master system. This data is saved locally on the master system. Due to the nature of the edge-based design this allows a user to configure multiple camera clusters across their property creating a smart connected security system that is easily expandable for a relatively low price, this cluster-based system will also allow users to toggle specific areas of their property depending on their circumstances. The end user will have graphical interface that will make it simple for them to review and manage their recorded security footage.
Location
School of Engineering & Computer Science
Start Date
4-5-2018 2:30 PM
End Date
4-5-2018 4:00 PM
SecureVis
School of Engineering & Computer Science
SecureVis is an open source smart computer vision driven security system that leverages an edge computing paradigm and deep learning computer vision algorithms to determine human activity across multiple spaces. Multi camera systems can be configured across various rooms / spaces, allowing for maximum coverage across a users property. Small camera clusters can be configured to communicate with a microcomputer (e.g. raspberry pi) which uses low computational cost OpenCV algorithms to determine if there is considerable movement within each cameras optical view. If such movement is detected data is passed over the network to a master system which utilizes a GPU accelerated Regional Convolutional Neural Network (e.g. YOLO) to determine if the subsequent activity is human. This human activity then triggers the affected cameras to stream security footage from their respective edge systems to the master system. This data is saved locally on the master system. Due to the nature of the edge-based design this allows a user to configure multiple camera clusters across their property creating a smart connected security system that is easily expandable for a relatively low price, this cluster-based system will also allow users to toggle specific areas of their property depending on their circumstances. The end user will have graphical interface that will make it simple for them to review and manage their recorded security footage.