US 11,709,553 B2
Automated prediction of a location of an object using machine learning
Raghuveer Prasad Nagar, Kota (IN); Kavita Prasad, Patna (IN); and Mahasweta Ghosh, Hooghly (IN)
Assigned to International Business Machines Corporation, Armonk, NY (US)
Filed by International Business Machines Corporation, Armonk, NY (US)
Filed on Feb. 25, 2021, as Appl. No. 17/185,560.
Prior Publication US 2022/0269349 A1, Aug. 25, 2022
This patent is subject to a terminal disclaimer.
Int. Cl. G06V 40/20 (2022.01); G06F 3/01 (2006.01); G06N 3/08 (2023.01); G06F 40/40 (2020.01); G06V 20/20 (2022.01)
CPC G06F 3/017 (2013.01) [G06F 40/40 (2020.01); G06N 3/08 (2013.01); G06V 20/20 (2022.01); G06V 40/20 (2022.01)] 15 Claims
OG exemplary drawing
 
1. A computer-implemented method comprising:
analyzing, using a machine learning model, a set of training data related to a plurality of body gestures performed by a user;
determining, using the machine learning model and based on the set of training data, a subset of body gestures that are associated with user movement indicating that the user is searching for one or more objects;
monitoring a data stream received from one or more observation devices;
detecting a body gesture initiated by the user from the data stream;
performing, using the machine learning model, a gesture analysis on the body gesture;
determining, based on the gesture analysis, that the body gesture indicates that the user is searching for an object;
identifying the object by analyzing a set of contextual data associated with the user, wherein the set of contextual data is at least one data type selected from the group of data types consisting of calendar data, text message communication data, and social media data, and wherein the set of contextual data comprises unstructured textual data related to performing an activity with the object;
predicting, in response to identifying the object, a location of the object by analyzing historic data associated with the object from the data stream; and
outputting the predicted location of the object to the user, in part, by activating a beam of light to physically identify the predicted location of the object.