Hyperzoom attribute analytics on the edge
A computer vision processor of a camera extracts attributes of persons or vehicles from hyperzooms generated from image frames. The hyperzooms represent traffic patterns. The extracting is performed using a feature extractor of an on-camera convolutional neural network (CNN) including an inverted residual structure. The attributes include at least colors of clothing of the persons or colors of the vehicles. Mobile semantic segmentation models of the CNN are generated using the hyperzooms and the attributes. Attribute analytics are generated by executing the mobile semantic segmentation models while obviating network usage by the camera. The attribute analytics are stored in a key-value database located on a memory card of the camera. A query is received from the server instance specifying one or more of the attributes. The attribute analytics are filtered using the one or more of the attributes to obtain a portion of the traffic patterns.
1. A computer vision method comprising:
extracting, by a computer vision processor, attributes of objects from hyperzooms generated by a first electronic device;
training a machine learning model using the hyperzooms and the attributes;
generating attribute analytics by executing the machine learning model on the attributes while obviating network usage by the first electronic device;
receiving a query specifying one or more of the attributes corresponding to a particular object;
in response to receiving the query, filtering the attribute analytics using the one or more of the attributes; and
transmitting the filtered attribute analytics to a second electronic device.
2. The method of claim 1 , wherein training the machine learning model comprises generating mobile semantic segmentation models using the hyperzooms and the attributes.
3. The method of claim 1 , wherein the query is a first query and the filtered attribute analytics are transmitted asynchronously with respect to a second query received from the second electronic device.
4. The method of claim 1 , comprising:
receiving a signal that a network connection to the first electronic device is active; and
responsive to receiving the signal, sending a descriptor of the first electronic device to the second electronic device.
5. The method of claim 4 , wherein the second electronic device is a server instance, the method comprising:
connecting to a communication channel generated by the server instance and referencing the descriptor, the communication channel for transmission of queries initiated by a user device.
6. The method of claim 1 , wherein the attribute analytics are stored in a key-value database as byte arrays on the first electronic device.
7. The method of claim 6 , wherein the key-value database stores the attribute analytics in a relational database management system (RDBMS).
8. A camera comprising:
a computer vision processor communicably coupled to the non-transitory computer-readable storage medium and configured to:
extract attributes of objects from hyperzooms captured by the camera;
train a machine learning model using the hyperzooms and the attributes;
generate attribute analytics by executing the machine learning model on the attributes while obviating network usage by the camera;
receive a query specifying one or more of the attributes corresponding to a particular object;
in response to receiving the query, filter the attribute analytics using the one or more of the attributes; and
transmit the filtered attribute analytics to an electronic device.
9. The camera of claim 8 , wherein the computer vision processor is configured to train the machine learning model by generating mobile semantic segmentation models using the hyperzooms and the attributes.
10. The camera of claim 8 , wherein the query is a first query and the filtered attribute analytics are transmitted asynchronously with respect to a second query received from the electronic device.
11. The camera of claim 8 , wherein the computer vision processor is configured to:
receive a signal that a network connection to the camera is active; and
responsive to receiving the signal, send a descriptor of the camera to the electronic device.
12. The camera of claim 8 , wherein the electronic device is a server instance, and the computer vision processor is configured to:
connect to a communication channel generated by the server instance and referencing the descriptor, the communication channel for transmission of queries initiated by a user device.
13. The camera of claim 8 , wherein the attribute analytics are stored in a key-value database as byte arrays on the first electronic device.
14. The camera of claim 8 , wherein the key-value database stores the attribute analytics in a relational database management system (RDBMS).
15. A system comprising:
one or more computer processors; and
a non-transitory computer-readable storage medium storing computer instructions, which when executed by the one or more computer processors cause the one or more computer processors to:
extract attributes of objects from hyperzooms generated by the system;
train a machine learning model using the hyperzooms and the attributes;
generate attribute analytics by executing the machine learning model on the attributes while obviating network usage by the system;
receive a query specifying one or more of the attributes corresponding to a particular object;
in response to receiving the query, filter the attribute analytics using the one or more of the attributes; and
transmit the filtered attribute analytics to an electronic device.
16. The system of claim 15 , wherein the computer instructions to train the machine learning model cause the one or more computer processors to generate mobile semantic segmentation models using the hyperzooms and the attributes.
17. The system of claim 15 , wherein the query is a first query and the filtered attribute analytics are transmitted asynchronously with respect to a second query received from the electronic device.
18. The system of claim 15 , wherein the computer instructions cause the one or more computer processors to:
receive a signal that a network connection to the system is active; and
responsive to receiving the signal, send a descriptor of the system to the electronic device.
19. The system of claim 15 , wherein the electronic device is a server instance, and the computer instructions cause the one or more computer processors to:
connect to a communication channel generated by the electronic device and referencing the descriptor, the communication channel for transmission of queries initiated by a user device.
20. The system of claim 15 , wherein the attribute analytics are stored in a key-value database as byte arrays on the first electronic device.