Granted Patent B2

US 12,561,971 · App. 18/136,184 · Granted Feb 24, 2026

Inventors: Grace Calvert Young (Mountain View, CA); Yangli Hector Yee (San Francisco, CA); Julia Black Ling (Menlo Park, CA); Cory Drew Schillaci (Berkeley, CA)

Assignee: TidalX AI Inc.

G06V20/05G06V10/25G06V10/764G06V20/70G06V2201/07

Abstract

Claims (55)

1 . An aquatic life data curation system comprising a plurality of computers and a plurality of storage devices storing instructions that are operable, when executed by the computers, to cause the computers to perform operations comprising:

obtaining aquatic life data;

identifying, from the aquatic life data, by a data annotation curator and in accordance with a set of data annotation rules, a first subset of the aquatic life data that is to be annotated by one or more data annotators to generate annotated aquatic life data and a second subset of the aquatic life data that is not to be annotated;

providing, by the data annotation curator, the first subset of the aquatic life data to the one or more data annotators, the one or more data annotators processing the first subset of the aquatic life data to generate the annotated aquatic life data;

providing, by the data annotation curator, the second subset of the aquatic life data to a data storage curator; and

determining, by the data storage curator and in accordance with a set of data storage rules, whether to store the second subset of the aquatic life data and, if so, which storage device in one or more storage devices to store the second subset of the aquatic life data.

2 . The aquatic life data curation system of claim 1 , wherein the aquatic life data comprises images that depict aquatic life captured by one or more vision sensors comprising a camera, a video camera, a thermal imaging sensor, or an infrared sensor.

3 . The aquatic life data curation system of claim 2 , wherein the aquatic life data comprises location or time metadata associated with the images.

4 . The aquatic life data curation system of claim 1 , wherein identifying the first subset of the aquatic life data that should be annotated by the one or more data annotators to generate the annotated aquatic life data comprises:

processing the aquatic life data to generate one or more data annotation curator outputs that characterize the aquatic life data;

identifying, based on the one or more data annotation curator outputs, a data annotator from among the one or more data annotators that have varying levels of annotation accuracy, throughput, capability, or a combination thereof; and

providing at least some of the aquatic life data to the identified data annotator.

5 . The aquatic life data curation system of claim 4 , wherein the data annotation curator outputs comprise one or more of: an image classification output, a video classification output, an object detection output, a clustering output, or an anomaly detection output.

6 . The aquatic life data curation system of claim 1 , wherein the annotated aquatic life data comprises:

data defining a plurality of bounding boxes in an aquatic life image; and

for each of the plurality of bounding boxes, a respective label defining that aquatic life belonging to an aquatic life category from a set of aquatic life categories is present in the region of an environment shown in the bounding box.

7 . The aquatic life data curation system of claim 1 , wherein the set of data annotation rules comprise:

a first data annotation rule specifying that aquatic life data which satisfies an information gain threshold should be annotated,

a second data annotation rule specifying that aquatic life data which satisfies an urgency threshold should be annotated, or

both the first and second data annotation rules.

8 . The aquatic life data curation system of claim 1 , wherein determining whether and if so which storage device in the one or more storage devices to store the second subset of the aquatic life data comprises:

identifying, based on the second subset of the aquatic life data, a storage device from among the one or more storage devices that (i) have varying levels of access speed, data security, or both, (ii) are in different geographic locations, or both (i) and (ii).

9 . The aquatic life data curation system of claim 8 , wherein the operations further comprise:

determining how the second subset of the aquatic life data should be stored in the identified storage device, comprising determining a number of copies, a resolution, or both of the second subset of the aquatic life data that should be stored.

10 . The aquatic life data curation system of claim 1 , wherein the operations further comprise:

applying data stored in the one or more storage devices as training data to train a machine learning model to process a model input that comprises an image depicting aquatic life and to generate a model output that characterizes the aquatic life depicted in the image.

11 . The aquatic life data curation system of claim 1 , wherein the set of data storage rules comprise one or more data compliance rules.

12 . One or more non-transitory computer-readable storage media storing instructions that when executed by one or more computers cause the one or more computers to perform operations comprising:

obtaining aquatic life data;

identifying, from the aquatic life data, by a data annotation curator and in accordance with a set of data annotation rules, a first subset of the aquatic life data that is to be annotated by one or more data annotators to generate annotated aquatic life data and a second subset of the aquatic life data that is not to be annotated;

providing, by the data annotation curator, the first subset of the aquatic life data to the one or more data annotators, the one or more data annotators processing the first subset of the aquatic life data to generate the annotated aquatic life data;

providing, by the data annotation curator, the second subset of the aquatic life data to a data storage curator; and

determining, by the data storage curator and in accordance with a set of data storage rules, whether to store the second subset of the aquatic life data and, if so, which storage device in one or more storage devices to store the second subset of the aquatic life data.

13 . A method performed by an aquatic life data curation system comprising a plurality of computers, wherein the method comprises:

obtaining aquatic life data;

identifying, from the aquatic life data, by a data annotation curator and in accordance with a set of data annotation rules, a first subset of the aquatic life data that is to be annotated by one or more data annotators to generate annotated aquatic life data and a second subset of the aquatic life data that is not to be annotated;

providing, by the data annotation curator, the first subset of the aquatic life data to the one or more data annotators, the one or more data annotators processing the first subset of the aquatic life data to generate the annotated aquatic life data;

providing, by the data annotation curator, the second subset of the aquatic life data to a data storage curator; and

determining, by the data storage curator and in accordance with a set of data storage rules, whether to store the second subset of the aquatic life data and, if so, which storage device in one or more storage devices to store the second subset of the aquatic life data.

14 . The method of claim 13 , wherein the aquatic life data comprises images that depict aquatic life captured by one or more vision sensors comprising a camera, a video camera, a thermal imaging sensor, or an infrared sensor.

15 . The method of claim 14 , wherein the aquatic life data comprises location or time metadata associated with the images.

16 . The method of claim 13 , wherein identifying the first subset of the aquatic life data that should be annotated by the one or more data annotators to generate the annotated aquatic life data comprises:

processing the aquatic life data to generate one or more data annotation curator outputs that characterize the aquatic life data;

identifying, based on the one or more data annotation curator outputs, a data annotator from among the one or more data annotators that have varying levels of annotation accuracy, throughput, capability, or a combination thereof; and

providing at least some of the aquatic life data to the identified data annotator.

17 . The method of claim 16 , wherein the data annotation curator outputs comprise one or more of: an image classification output, a video classification output, an object detection output, a clustering output, or an anomaly detection output.

18 . The method of claim 13 , wherein the annotated aquatic life data comprises:

data defining a plurality of bounding boxes in an aquatic life image; and

for each of the plurality of bounding boxes, a respective label defining that aquatic life belonging to an aquatic life category from a set of aquatic life categories is present in the region of an environment shown in the bounding box.

19 . The method of claim 13 , wherein the set of data annotation rules comprise:

a first data annotation rule specifying that aquatic life data which satisfies an information gain threshold should be annotated,

a second data annotation rule specifying that aquatic life data which satisfies an urgency threshold should be annotated, or

both the first and second data annotation rules.

20 . The method of claim 13 , wherein determining whether and if so which storage device in the one or more storage devices to store the second subset of the aquatic life data comprises:

identifying, based on the second subset of the aquatic life data, a storage device from among the one or more storage devices that (i) have varying levels of access speed, data security, or both, (ii) are in different geographic locations, or both (i) and (ii).

Assignments (2)

Continuity (1)

References Cited (15)