US 12,482,245 B2
System and method for personalizing a computer vision model based on use cases
Job Varughese Philip, Mumbai (IN); Harshit Bhatia, New Delhi (IN); and Happy Mishra, Vadodara (IN)
Assigned to ASSERT SECURE TECH PVT. LIMITED, Mumbai (IN)
Filed by ASSERT SECURE TECH PVT. LIMITED, Mumbai (IN)
Filed on Jan. 31, 2025, as Appl. No. 19/043,073.
Claims priority of application No. 202421006710 (IN), filed on Feb. 1, 2024.
Prior Publication US 2025/0252720 A1, Aug. 7, 2025
Int. Cl. G06V 10/778 (2022.01); G06V 10/26 (2022.01)
CPC G06V 10/7788 (2022.01) [G06V 10/26 (2022.01)] 10 Claims
OG exemplary drawing
 
1. A method for personalizing a computer vision (CV) model based on use cases, wherein the method comprises:
segmenting, videos received from a user device associated with a plurality of users into a sequence of individual frames, wherein the videos are at least one of default videos, user-uploaded videos, or the videos captured using provided camera details;
extracting, a first frame from the sequence of individual frames to initiate a computer vision model process, wherein the first frame is configured to enable tracking of coordinates based on user inputs;
generating an event in response to user interaction with the first frame, wherein the user interaction comprises one or more user inputs performed using one or more input devices;
generating a plurality of preference segments to a plurality of user interfaces of the user device to enable the user to select a preference to initiate the use cases, wherein the user is enabled to select at least one of a first use case, or a second use case from among the plurality of preference segments;
configuring corresponding computer vision (CV) models personalized for each selected use case in parallel by processing the selection the first use case and the second use case;
tracking the coordinates within the first frame based on the one or more user inputs using the one or more input devices based on the configured CV model;
converting the one or more user inputs performed within the first frame using the one or more input devices into the coordinates, wherein the coordinates are stored in a database in a predefined format;
processing the coordinates stored in the database into a corresponding CV model, wherein the database comprises a first module and a second module;
allowing the user to provide an input, through the one or more input devices to accept or discard a change in the first frame;
defining a plurality of parameters comprising at least one of a region of interest or a line control in the first frame; and
executing the first use case and the second use case by utilizing at least one of the default models associated with the system, or generated user-specific CV models, wherein the generated user-specific CV models are dynamically configured to generate a new CV model based on user-defined parameters.