Training model to identify items based on image data and load curve data

Abstract

20 26 20 53 55 07 J ul 2 02 6 A B S T R A C T 2 0 2 6 2 0 5 3 5 5 0 7 J u l 2 0 2 6

Claims

1. A method comprising:at a computing system comprising a processor and a memory:receiving load data from a plurality of load sensors, wherein each load sensor is coupled to a different location of a storage area of a shopping cart, and wherein the load data received from each load sensor comprises a load measurement captured by the load sensor at each of a series of timestamps;detecting that an item was added to the storage area of the shopping cart during one or more of the series of timestamps of the load data;identifying a set of load measurements captured by one or more of the plurality of load sensors during the one or more timestamps when the item was added, the identified set of load measurements comprising a load curve for the corresponding load sensor of the plurality of load sensors;applying an item recognition model to a load curve corresponding to each load sensor of the plurality of load sensors to generate an item identifier prediction for the item, wherein the item recognition model is a machine-learning model that is trained to identify items based on load curve information associated with a plurality of load sensors; andstoring the generated item identifier prediction.

2. The method of claim 1, further comprising:generating, using the item recognition model, a predicted identification of the item; andproviding, for rendering at a display system associated with the shopping cart, information about the predicted identification of the item.2026205355   07 Jul 20263. The method of claim 1, further comprising:identifying a set of timestamps that are most likely to correspond to the item being placed in the storage area of the shopping cart by applying a timestamp detection model to the load data from each of the plurality of load sensors, wherein the timestamp detection model is a machine-learning model trained to identify timestamps that correspond to an item being added to a storage area of a shopping cart.

4. The method of claim 1, further comprising training the item recognition model based on a set of training examples of items placed in the shopping cart.

5. The method of claim 1, further comprising:receiving aggregate image data from a plurality of cameras, wherein each camera is coupled to a different location of the storage area of the shopping cart, and wherein the image data received from each camera comprises an image frame captured at each of the series of timestamps; andwherein applying the item recognition model further comprises applying the item recognition model to the aggregate image data.

6. The method of claim 1, wherein the plurality of load sensors comprises a plurality of scales.

7. The method of claim 1, wherein the plurality of load sensors comprises fourscales, each scale affixed substantially near the one of four corners of the shopping cart.

8. A non-transitory computer-readable storage medium storing executable instructions that, when executed by a hardware processor, cause the hardware processor to perform steps comprising:2026205355   07 Jul 2026receiving load data from a plurality of load sensors, wherein each load sensor is coupled to a different location of a storage area of a shopping cart, wherein the load data received from each load sensor comprises a load measurement captured by the load sensor at each of a series of timestamps;detecting that an item was added to the storage area of the shopping cart during one or more of the series of timestamps of the load data;identifying a set of load measurements captured by one or more of the plurality of load sensors during the one or more timestamps when the item was added, the identified set of load measurements comprising a load curve for the corresponding load sensor of the plurality of load sensors;applying an item recognition model to a load curve corresponding to each load sensor of the plurality of load sensors to generate an item identifier prediction for the item, wherein the item recognition model is a machine-learning model that is trained to identify items based on load curve information associated with a plurality of load sensors; andstoring the generated item identifier prediction.

9. The non-transitory computer-readable storage medium of claim 8, the steps further comprising:generating, using the item recognition model, a predicted identification of the item; andproviding, for rendering at a display system associated with the shopping cart, information about the predicted identification of the item.

10. The non-transitory computer-readable storage medium of claim 8, the steps further comprising:2026205355   07 Jul 2026identifying a set of timestamps that are most likely to correspond to the item being placed in the storage area of the shopping cart by applying a timestamp detection model to the load data from each of the plurality of load sensors, wherein the timestamp detection model is a machine-learning model trained to identify timestamps that correspond to an item being added to a storage area of a shopping cart.

11. The non-transitory computer-readable storage medium of claim 8, the steps further comprising training the item recognition model based on a set of training examples of items placed in the shopping cart.

12. The non-transitory computer-readable storage medium of claim 8, the steps further comprising:receiving aggregate image data from a plurality of cameras, wherein each camera is coupled to a different location of the storage area of the shopping cart, wherein the image data received from each camera comprises an image frame captured at each of the series of timestamps; andwherein applying the item recognition model further comprises applying the item recognition model to the aggregate image data.

13. The non-transitory computer-readable storage medium of claim 8, wherein the plurality of load sensors comprises a plurality of scales.

14. The non-transitory computer-readable storage medium of claim 8, wherein the plurality of load sensors comprises four scales, each scale affixed substantially near the one of four corners of the shopping cart.

15. A computer system comprising:2026205355   07 Jul 2026a hardware processor; anda non-transitory computer-readable storage medium storing executable instructions that, when executed, cause the hardware processor to perform steps comprising:receiving load data from a plurality of load sensors, wherein each load sensor is coupled to a different location of a storage area of a shopping cart, wherein the load data received from each load sensor comprises a load measurement captured by the load sensor at each of a series of timestamps;detecting that an item was added to the storage area of the shopping cart during one or more of the series of timestamps of the load data;identifying a set of load measurements captured by one or more of the plurality of load sensors during the one or more timestamps when the item was added, the identified set of load measurements comprising a load curve for the corresponding load sensor of the plurality of load sensors;applying an item recognition model to a load curve corresponding to each load sensor of the plurality of load sensors to generate an item identifier prediction for the item, wherein the item recognition model is a machine-learning model that is trained to identify items based on load curve information associated with a plurality of load sensors; andstoring the generated item identifier prediction.

16. The computer system of claim 15, the steps further comprising:generating, using the item recognition model, a predicted identification of the item;and2026205355   07 Jul 2026providing, for rendering at a display system associated with the shopping cart, information about the predicted identification of the item.

17. The computer system of claim 15, the steps further comprising:identifying a set of timestamps that are most likely to correspond to the item being placed in the storage area of the shopping cart by applying a timestamp detection model to the load data from each of the plurality of load sensors, wherein the timestamp detection model is a machine-learning model trained to identify timestamps that correspond to an item being added to a storage area of a shopping cart.

18. The computer system of claim 15, the steps further comprising training the item recognition model based on a set of training examples of items placed in the shopping cart.

19. The computer system of claim 15, the steps further comprising:receiving aggregate image data from a plurality of cameras, wherein each camera is coupled to a different location of the storage area of the shopping cart, wherein the image data received from each camera comprises an image frame captured at each of the series of timestamps; andwherein applying the item recognition model further comprises applying the item recognition model to the aggregate image data.

20. The computer system of claim 15, wherein the plurality of load sensors comprises four scales, each scale affixed substantially near the one of four corners of the shopping cart.

Images