Adjusting areas of interest for motion detection in camera scenes

Patent metadata

Field	Value
Publication number	US-12008814-B2
Application number	US-202217739237-A
Country	US
Kind code	B2
Filing date	May 9, 2022
Priority date	May 11, 2021
Publication date	Jun 11, 2024
Grant date	Jun 11, 2024

Abstract

Disclosed are methods, systems, and apparatus for adjusting areas of interest for motion detection in camera scenes. A method includes obtaining a map of false motion event detections using a first area of interest; identifying an overlap area between the map of false detections and the first area of interest; determining a second area of interest that includes portions of the first area of interest and excludes at least a part of the overlap area; obtaining a map of true motion event detections using the first area of interest; determining whether true detections using the second area of interest compared to true detections using the first area of interest satisfies performance criteria; and in response to determining that true detections using the second area of interest compared to true detections using the first area of interest satisfies performance criteria, providing the second area of interest for use in detecting events.

First claim

What is claimed is:

1. A method comprising:
providing, for use in detecting events depicted in one or more first images, a first area of interest of a scene within a field of view of a camera;
obtaining a map of false motion events that were detected using the first area of interest;
in response to obtaining the map of false motion events, identifying an overlap area between the map of false motion events and the first area of interest;
determining a second area of interest that includes one or more portions of the first area of interest and excludes at least a part of the overlap area between the map of false motion events and the first area of interest;
obtaining a map of true motion events that were detected using the first area of interest;
determining, using the map of true motion events, a performance metric for using the first area of interest to detect events;
determining, using the map of true motion events, a predicted performance metric for using the second area of interest to detect events;
determining whether the performance metric for using the first area of interest to detect events and the predicted performance metric for using the second area of interest to detect events satisfy performance criteria; and
in response to determining that the performance metric for using the first area of interest to detect events and the predicted performance metric for using the second area of interest to detect events satisfy the performance criteria, providing the second area of interest for use in detecting events depicted in one or more different, second images instead of the first area of interest.

2. The method of claim 1 , comprising generating the map of false motion events and the map of true motion events by:
obtaining data defining the first area of interest;
obtaining the one or more first images;
detecting a plurality of motion events in the one or more first images, each motion event representing motion of an object within the field of view of the camera; and
classifying each motion event as a false motion event or a true motion event.

3. The method of claim 2 , comprising:
determining a traversal score for each motion event, wherein the traversal score indicates an extent of motion of the object across the field of view of the camera,
wherein classifying each motion event as a true motion event or a false motion event uses the traversal score.

4. The method of claim 3 , wherein determining a traversal score comprises, for each motion event representing motion of an object:
determining a centroid location of the object in each frame of a frame set, the frame set including multiple sequential image frames in which the object was detected during the motion event;
determining a traversal metric using a distance between the centroid location in a first frame of the frame set and the centroid location in a final frame of the frame set; and
determining, for each motion event, a corresponding traversal score by combining traversal metrics for a plurality of frame sets, each frame set including multiple sequential image frames in which the object was detected during the motion event.

5. The method of claim 1 , wherein the map of false motion events includes an outline encompassing pixels that correspond to locations of the scene where false motion events occurred.

6. The method of claim 1 , wherein the map of true motion events includes an outline encompassing pixels that correspond to locations of the scene where true motion events occurred, the method comprising generating the map of true motion events by:
generating a plurality of bounding boxes, wherein each bounding box includes at least an upper boundary and a lower boundary and encloses a location of the scene where a true motion event occurred; and
generating the outline, wherein the outline encompasses a portion of each bounding box that includes the lower boundary of the bounding box.

7. The method of claim 1 , wherein determining the second area of interest comprises:
labeling segments of the scene according to objects represented by the segments;
generating a copy of the first area of interest; and
adjusting the copy of the first area of interest based on the labels of the segments of the scene to obtain the second area of interest.

8. The method of claim 7 , wherein adjusting the copy of the first area of interest based on the labels of the segments of the scene comprises:
determining a label for a particular segment of the scene within the overlap area between the map of false motion events and the first area of interest;
classifying the label of the particular segment as corresponding to an object that is not of interest; and
based on classifying the label of the particular segment as corresponding to an object that is not of interest, removing the particular segment from the copy of the first area of interest.

9. The method of claim 7 , wherein adjusting the copy of the first area of interest using the labels of the segments of the scene comprises:
determining a label for a particular segment of the scene within the overlap area between the map of false motion events and the first area of interest;
classifying the label of the particular segment as corresponding to an object that is of interest; and
based on classifying the label of the particular segment as corresponding to an object that is of interest, maintaining the particular segment within the first area of interest.

10. The method of claim 1 , wherein determining whether the performance metric for using the first area of interest to detect events and the predicted performance metric for using the second area of interest to detect events satisfy performance criteria comprises determining an impact of using the second area of interest on at least one of:
a recall of the camera;
a detection latency of the camera; or
a degree of overlap between the respective area of interest and the map of true motion events.

11. The method of claim 1 , wherein the performance criteria comprise a maximum threshold impact on a recall of the camera, the recall of the camera using a ratio of a number of true motion events to a total number of true objects of interest depicted in images captured by the camera during a duration of time.

12. The method of claim 1 , wherein the performance criteria comprise a maximum threshold impact on a true motion overlap score of the camera, the true motion overlap score using a degree of overlap between a map of true motion events and a respective area of interest.

13. The method of claim 1 , wherein the performance criteria comprise a maximum threshold impact on detection latency of the camera, the detection latency using a time delay between a time when an event occurs and a time when the camera detects the event.

14. The method of claim 1 comprising:
detecting a decrease in performance of the camera while detecting events using the second area of interest; and
in response to detecting the decrease in performance of the camera, providing the first area of interest for use in detecting events.

15. The method of claim 14 , wherein detecting the decrease in performance comprises determining that an average rate of false motion events that occur while detecting events using the second area of interest is greater than an average rate of false motion events that occur while detecting events using the first area of interest.

16. A system comprising one or more computers and one or more computer storage media storing instructions that are operable, when executed by the one or more computers, to cause the one or mor

Assignees

• Objectvideo Labs Llc

Inventors

• Ramanathan Narayanan
• Qian Gang
• Carmena Eduardo Romera
• Madden Donald Gerard
• Beach Allison

Classifications

• G06T7/20

  Analysis of motion (motion estimation for coding, decoding, compressing or decompressing digital video signals H04N19/43, H04N19/51) · CPC title

• G06T2207/10016

  Video; Image sequence · CPC title

• H04N7/188

  Capturing isolated or intermittent images triggered by the occurrence of a predetermined event, e.g. an object reaching a predetermined position (signal generation from motion picture films H04N5/253) · CPC title

• G06T2207/30232

  Surveillance · CPC title

• G08B13/19602

  Image analysis to detect motion of the intruder, e.g. by frame subtraction · CPC title