Household appliance video analysis
A video captured by a camera assembly of a household appliance includes a plurality of frames. Methods of analyzing the video may include calculating a plurality of median frames of the video, determining an area of one of the plurality of frames contains an object of interest, and isolating the determined area for analysis of the object of interest. Methods of analyzing the video may include identifying an object of interest in a minimum number of consecutive frames of the video, determining a motion vector of the object of interest based on the consecutive frames of the video, and comparing the motion vector with predetermined in and out vectors to determine whether the object of interest was added to or removed from the household appliance. Such methods may also include adding or removing the object of interest to or from a virtual inventory of the household appliance.
1 . A method of analyzing a video captured by a camera assembly of a household appliance, the video comprising a plurality of frames, the method comprising:
calculating a plurality of median frames of the video, wherein each median frame is a median frame for a predetermined number of frames of the plurality of frames;
calculating frame differences, wherein calculating frame differences comprises subtracting each of the predetermined number of frames from the respective median frame;
creating a subtracted image from each of the predetermined number of frames and the respective median frame based on the calculated frame differences;
creating a smoothed mask from the subtracted image;
determining an area of one of the plurality of frames contains an object of interest; and
isolating the determined area for analysis of the object of interest.
2 . The method of claim 1 , further comprising comparing a masked portion of the smoothed mask to a predetermined area threshold, and wherein determining the area of one of the plurality of frames contains the object of interest comprises determining the masked portion is greater than the predetermined area threshold.
3 . The method of claim 1 , wherein isolating the determined area for analysis of the object of interest comprises transmitting the determined area to an artificial intelligence model for object classification and location.
4 . The method of claim 1 , wherein the analysis of the object of interest comprises determining a motion vector of the object of interest.
5 . The method of claim 4 , further comprising comparing the motion vector with predetermined in and out vectors to determine whether the object of interest was added to or removed from the household appliance.
6 . The method of claim 5 , further comprising adding the object of interest to a virtual inventory of the household appliance in response to determining the object of interest was added to the household appliance or removing the object of interest from the virtual inventory in response to determining the object of interest was removed from the household appliance.
7 . The method of claim 4 , wherein the analysis of the object of interest further comprises identifying the object of interest in a minimum number of consecutive frames of the video, wherein the motion vector is determined based on the consecutive frames of the video.
8 . A method of analyzing a video captured by a camera assembly of a household appliance, the video comprising a plurality of frames, the method comprising:
identifying an object of interest in a minimum number of consecutive frames of the video;
determining a motion vector of the object of interest based on the consecutive frames of the video;
comparing the motion vector with predetermined in and out vectors to determine whether the object of interest was added to or removed from the household appliance;
calculating a plurality of median frames of the video, wherein each median frame is a median frame for a predetermined number of frames of the plurality of frames, and wherein the object of interest is identified based on the plurality of median frames of the video;
calculating frame differences by subtracting each of the predetermined number of frames from the respective median frame;
creating a subtracted image from each of the predetermined number of frames and the respective median frame based on the calculated frame differences;
creating a smoothed mask from the subtracted image; and
updating a virtual inventory of the household appliance in response to determining whether the object of interest was added to or removed from the household appliance.
9 . The method of claim 8 , further comprising comparing a masked portion of the smoothed mask to a predetermined area threshold, and wherein identifying the object of interest comprises determining the masked portion is greater than the predetermined area threshold, wherein the object of interest is depicted within the masked portion.
10 . The method of claim 9 , further comprising classifying and locating the object of interest using an artificial intelligence model.