Aerial surveillance for premises security systems
A method implemented by a system for security comprising an unmanned aerial vehicle (UAV) and an analytics device configured to communicate with a remote monitoring system and the UAV is provided. Media data from the UAV is received at an analytics device, where the media data includes surveillance information corresponding to a premises under surveillance. Security attributes associated with the premises are detected based at least in part on a first level of machine learning (ML) analysis performed on the media data. The media data are transmitted by the analytics device to the remote monitoring system for a second level of ML analysis based at least in part on the security attribute, where the first level of ML analysis is less computationally expensive compared to the second level of ML analysis, and the second level of ML analysis is performed at the remote monitoring system on the media data.
1. An analytics device for a media source, the analytics device being configured to communicate with a remote monitoring system and the media source, the analytics device comprising processing circuitry configured to:
receive data from the media source, the data comprising surveillance information corresponding to a premises;
detect a security attribute associated with the premises based at least in part on a first level of machine learning (ML) analysis of the data; and
in response to the security attribute, cause transmission of at least a portion of the data to the remote monitoring system for a second level of ML analysis, the first level of ML analysis being different from the second level of ML analysis.
2. The analytics device of claim 1 , wherein the processing circuitry is further configured to:
determine an alarm probability level associated with the security attribute; and
transmit at least the portion of the data to the remote monitoring system only when the alarm probability level exceeds a preconfigured threshold.
3. The analytics device of claim 1 , wherein the media source comprises an unmanned aerial vehicle (UAV) and the processing circuitry is further configured to:
receive a security alert from the remote monitoring system indicating a premises area associated with the premises;
responsive to receiving the security alert, cause transmission of an indication to the UAV for the UAV to transit to the premises area; and
the receiving of the data being responsive to the indication, the data including surveillance information associated with the premises area.
4. The analytics device of claim 1 , wherein the media source comprises an unmanned aerial vehicle (UAV), and the data comprises at least one of:
a video surveillance feed recorded by a camera of the UAV;
biometric sensor information recorded by the UAV;
fire sensor information recorded by the UAV; or
gas sensor information recorded by the UAV.
5. The analytics device of claim 1 , wherein the data is media data; and
the processing circuitry is further configured to detect the security attribute of the media data by:
identifying a dominant object in an image frame of the media data;
determining a classification of the dominant object based on the first level of ML analysis; and
mapping the classification to the security attribute based on a preconfigured security attribute mapping.
6. The analytics device of claim 1 , wherein the security attribute comprises at least one of:
an alarm probability level;
a threat severity level indication;
a priority level indication;
a broken window indication;
an open door indication;
an intruder indication; or
an identification of a person in a premises area.
7. The analytics device of claim 1 , wherein the analytics device lacks a configuration for notifying first responders and the remote monitoring system has a configuration for notifying first responders based on the second level of ML analysis.
8. The analytic device of claim 1 , wherein the first level of ML analysis at least one of:
uses a less complex ML model compared to an ML model used in the second level of ML analysis; or
requires less computational resources than the second level of ML analysis.
9. A system, comprising:
an analytics device being configured to communicate with a remote monitoring system and a media source, the analytics device comprising processing circuitry configured to:
receive data from the media source, the data comprising surveillance information corresponding to a premises;
detect a security attribute associated with the premises based at least in part on a first level of machine learning (ML) analysis of the data; and
in response to the security attribute, cause transmission of at least a portion of the data to the remote monitoring system for a second level of ML analysis, the first level of ML analysis being different from the second level of ML analysis; and
the remote monitoring system comprising processing circuitry configured to:
receive at least the portion of the data from the analytics device; and
responsive to receiving at least the portion of the data, perform the second level of ML analysis on at least the portion of the data.
10. The system of claim 9 , wherein the processing circuitry of the analytics device is further configured to:
determine an alarm probability level associated with the security attribute; and
transmit at least the portion of the data to the remote monitoring system only when the alarm probability level exceeds a preconfigured threshold.
11. The system of claim 9 , wherein the processing circuitry of the remote monitoring system is further configured to receive, from the analytics device, an indication of the security attribute; and
the second level of ML analysis comprises determining a prediction accuracy of the security attribute.
12. The system of claim 11 , wherein the processing circuitry of the remote monitoring system is further configured to:
cause transmission of an emergency alert to first responders only when the prediction accuracy exceeds a preconfigured threshold.
13. The system of claim 9 , wherein the second level of ML analysis comprises detecting an additional security attribute associated with the premises based at least in part on the data; and
the processing circuitry of the remote monitoring system is further configured to cause transmission of an emergency alert to first responders based at least in part on the additional security attribute.
14. The system of claim 13 , wherein the processing circuitry of the remote monitoring system is further configured to detect the additional security attribute by:
identifying a dominant object in an image frame of the data;
determining a classification of the dominant object; and
mapping the classification to the additional security attribute based on a preconfigured security attribute mapping.
15. The system of claim 9 , wherein the media source comprises an unmanned aerial vehicle (UAV);
the processing circuitry of the remote monitoring system is further configured to:
receive security information from at least one of a premises security system or a remote first responder service; and
cause transmission of a security alert to the analytics device based at least in part on the security information, the security alert indicating a premises area associated with the premises for monitoring; and
the processing circuitry of the analytics device is further configured to:
receive the security alert from the remote monitoring system indicating the premises area of the premises for monitoring;
responsive to receiving the security alert, cause transmission of an indication to the UAV for the UAV to transit to the premises area; and
the receiving of the data being responsive to the indication, the data including surveillance information associated with the premises area.
16. The system of claim 9 , wherein the media source comprises an unmanned aerial vehicle (UAV), and the data comprises at least one of:
a video surveillance feed recorded by a camera of the UAV;
biometric sensor information recorded by the UAV;
fire sensor information recorded by the UAV; or
gas sensor information recorded by the UAV.
17. The system of claim 9 , wherein the security attribute comprises at least one of:
an alarm probability level;
a threat severity level indication;
a priority level indication;
a broken window indication;
an open door indication;
an intruder indication; or
an identification of a person in a premises area.
18. A method implemented by an analytics device for a media source, the analytics device being configured to communicate with a remote monitoring system and the media source, the method comprising:
receiving data from the media source, the data comprising surveillance information corresponding to a premises;
detecting a security attribute associated with the premises based at least in part on a first level of machine learning (ML) analysis of the data; and
in response to the security attribute, transmitting at least a portion of the data to the remote monitoring system for a second level of ML analysis, the first level of ML analysis being different from the second level of ML analysis.
19. The method of claim 18 , wherein the media source comprises an unmanned aerial vehicle (UAV); and
the method further comprises:
receiving a security alert from the remote monitoring system indicating a premises area associated with the premises;
responsive to receiving the security alert, transmitting an indication to the UAV for the UAV to transit to the premises area; and
the receiving of the data being responsive to the indication, the data including surveillance information associated with the premises area.
20. The method of claim 18 , wherein the first level of ML analysis at least one of:
uses a less complex ML model compared to an ML model used in the second level of ML analysis; or
requires less computational resources than the second level of ML analysis.