US 12,444,027 B2
Processes to create a docu-narrative, including evidentiary quality images and evidentiary objects within the images, and an image chain of custody, with multi-level authentication and certification
Juan Navas, Fair Lawn, NJ (US); and Brian Cury, Alpine, NJ (US)
Assigned to EarthCam Inc., Upper Saddle River, NJ (US)
Filed by EarthCam Inc., Upper Saddle River, NJ (US)
Filed on May 20, 2023, as Appl. No. 18/199,949.
Prior Publication US 2024/0386530 A1, Nov. 21, 2024
Int. Cl. G06T 5/70 (2024.01); G06T 5/50 (2006.01); G06T 5/77 (2024.01)
CPC G06T 5/70 (2024.01) [G06T 5/50 (2013.01); G06T 5/77 (2024.01)] 36 Claims
OG exemplary drawing
 
1. A method for producing a visual record of events that occur at a location, the method comprising:
determining a time duration to be covered by the visual record;
determining a number of images to be included in the visual record to satisfy the time duration;
determining a hardware configuration for an image acquisition device;
implementing the hardware configuration on the image acquisition device;
identifying a benchmark resolution and focus to be used for one or more images in the visual record;
employing authentication, certification or verification procedures to establish and maintain an evidentiary chain of custody for the visual record;
authenticating, certifying, or verifying that one or more images within the visual record have not been altered, concealed, falsified, or destroyed;
selecting a sample image from the visual record;
determining a quality of the sample image based on one or more of a mean average precision score of the sample image, orientation of the sample image, and an aspect ratio of the sample image, wherein the mean average precision score is indicative of differences between an image of an object and an image of a ground truth object;
creating a base model comprising one or more characteristics of a graphics processing unit, the characteristics comprising image resolution, speed of inference or processing of each image within the visual record, and accuracy of the images, wherein accuracy is indicative of differences between an image of an object and an image of the ground truth object;
modifying characteristics of an image such that an image in the visual record appears substantially consistent with a previous and a next image within the visual record;
identifying a target object within the visual record;
creating a target object model for use in identifying the target object within an image, wherein the target model comprises characteristics of the target object;
developing target object detection specifications;
determining whether use of the target object detection specifications to detect target objects within an image yields an acceptable detection score based on a predetermined rubric; and
determining whether a predetermined percentage of target objects in the plurality of images have been correctly identified.