Granted Patent B2

US 12,541,944 · App. 18/636,049 · Granted Feb 3, 2026

Inventors: Narasimhachary Nallana Chakravarty (Rollinsford, NH); Guohua Min (Exeter, NH); Edward L. Hill (Conway, NH); Brett Bilbrey (Sunnyvale, CA)

Assignee: Position Imaging, Inc.

G06V10/454G06T7/20G06V10/803G06V10/82G06V20/10G06V20/52G06T2207/20081G06T2207/20084G06V10/62

Abstract

Claims (39)

1 . A method of identifying and tracking objects, the method comprising the steps of:

capturing a plurality of images of an area designated for holding objects;

submitting the plurality of images to a processor;

performing, by the processor, a change analysis operation to determine if changes in a state of the plurality of images are detected;

determining whether a neural network detects an object in the area captured in the plurality of images;

executing a deep neural network re-training decision making process, including:

performing an action that is executed in response to (1) in an image of the plurality of images an actual movement or non-movement of the object in a region of the image or the area designated for holding objects, (2) a determination by the change analysis operation whether changes in the state of the plurality of images are detected, and (3) a determination by the deep neural network whether the object is detected in the region of the image or the area designated for holding objects.

2 . The method of claim 1 , wherein the action includes retraining the deep neural network.

3 . The method of claim 1 , the action further comprising:

applying an external input to recognize the object;

associating object information with the image; and

retraining the deep neural network.

4 . The method of claim 3 , wherein the object information includes a label.

5 . The method of claim 3 , wherein the external input includes a user interface.

6 . The method of claim 2 , wherein the action includes:

automatically detecting label information of the object;

associating the label information with the image; and

processing the label information and the image to retrain the deep neural network.

7 . The method of claim 6 , wherein the label information of the object includes a combination of location, type, size, shape, and color of the object.

8 . The method of claim 1 , further comprising:

tracking a location of the object after the object is detected while the object is in the area designated for holding objects.

9 . The method of claim 1 , wherein if (3) the deep neural network fails to determine that the object is detected in the region of the image or the area designated for holding objects, then the region of the image or the area is cropped and the image is returned to (3) a determining by the deep neural network whether the object is detected in the cropped image.

10 . A system for identifying and tracking objects, comprising:

a camera having a field of view that captures a plurality of images in the field of view of the camera;

a change detector that detects changes in the plurality of images; and

a deep neural network (DNN) that is trained to detect objects in the plurality of images,

wherein the DNN is retrained if a determination is made that in an image of the plurality of images there is an actual movement of an object of the detected objects, that a change in the plurality of images is detected, and that the deep neural network detects the object regardless of whether the object is recognized.

11 . The system of claim 10 , further comprising

applying an external input to recognize the object;

associating object information with the image; and

retraining the deep neural network.

12 . The system of claim 11 , wherein the object information includes a label.

13 . The system of claim 11 , wherein the external input includes a user interface.

14 . The system of claim 10 , further comprising:

automatically detecting label information of the object;

associating the label information with the image; and

processing the label information and the image to retrain the deep neural network.

15 . The system of claim 14 , wherein the label information of the object includes a combination of location, type, size, shape, and color of the object.

16 . The system of claim 10 , wherein if the deep neural network fails to determine that the object is detected in a region of the image or an area designated for holding objects, then the region of the image or the area is cropped and the deep neural network determines whether the object is detected in the cropped image.

Assignments (3)

Continuity (4)

References Cited (400)