Item recognition enhancements

1 . A system comprising:

a scanner configured to scan an identifier on a first item;

a camera configured to capture a first image of the first item as the scanner is scanning the identifier on the first item;

a memory; and

one or more processors communicatively coupled to the memory, a combination of the one or more processors configured to:

obtain a scanned identity of the first item based, at least in part, on the scanner scanning the identifier on the first item;

extract information about the first item from the first image of the first item;

determine, based on the information about the first item, a first probability for a first identity of the first item and a second probability for a second identity of the first item;

in response to determining that both the first probability and the second probability are below a threshold, determine, based on a shopping history of a user purchasing the first item, that the user previously purchased a second item comprising a characteristic;

apply a first weight to the first probability based on the characteristic to produce a first weighted probability;

apply a second weight to the second probability based on the characteristic to produce a second weighted probability;

assign the first identity as a predicted identity of the first item based on the first weighted probability and the second weighted probability;

determine a mismatch between the scanned identity of the first item and the predicted identity of the first item;

generate an alert indicating the mismatch between the scanned identity of the first item and the predicted identity of the first item; and

display the alert on a display device for viewing by one or more persons.

2 . The system of claim 1 , wherein applying the first weight to the first probability is based on the first identity having the characteristic.

3 . The system of claim 2 , wherein applying the second weight to the second probability is based on the second identity lacking the characteristic.

4 . The system of claim 1 , wherein assigning the first identity as the predicted identity of the first item is in response to determining that (i) the first weighted probability exceeds the threshold and (ii) the second weighted probability falls below the threshold.

5 . The system of claim 1 , wherein the combination of the one or more processors is further configured to:

determine, based on the first image of the first item, a third probability for a third identity of the first item;

apply a third weight to the third probability based on the characteristic to produce a third weighted probability; and

in response to determining that both the first weighted probability and the third weighted probability exceed the threshold, prompt the user to select the first identity or the third identity, wherein assigning the first identity as the predicted identity of the first item is further based on the user selecting the first identity rather than the third identity.

6 . The system of claim 5 , wherein prompting the user comprises presenting, to the user, a second image for the first identity and a third image for the third identity.

7 . The system of claim 6 , wherein the first image and the third image are presented in an order based on the first weighted probability and the third weighted probability.

8 . The system of claim 5 , wherein prompting the user further comprises refraining from presenting, to the user, a fourth image for the second identity in response to determining that the second weighted probability falls below the threshold.

9 . The system of claim 1 , wherein assigning the first identity as the predicted identity of the first item is further based on at least one of a weight of the first item, an inventory count, or an expiration date.

10 . The system of claim 1 , wherein applying the second weight to the second probability is further based on the shopping history indicating that the user returned the second item.

11 . A method comprising:

obtaining, by one or more processors, a scanned identity of a first item based, at least in part, on scanning an identifier on the first item;

capturing, by a camera, a first image of the first item as the identifier on the first item is being scanned;

extracting, by the one or more processors, information about the first item from the first image;

determining, by the one or more processors and based on the information about the first item, a first probability for a first identity of the first item and a second probability for a second identity of the first item;

in response to determining that both the first probability and the second probability are below a threshold, determining, by the one or more processors and based on a shopping history of a user purchasing the first item, that the user purchasing the first item previously purchased a second item comprising a characteristic;

applying, by the one or more processors, a first weight to the first probability based on the characteristic to produce a first weighted probability;

applying, by the one or more processors, a second weight to the second probability based on the characteristic to produce a second weighted probability;

assigning, by the one or more processors, the first identity as a predicted identity of the first item based on the first weighted probability and the second weighted probability;

determining, by the one or more processors, a mismatch between the scanned identity of the first item and the predicted identity of the first item;

generating, by the one or more processors, an alert indicating the mismatch between the scanned identity of the first item and the predicted identity of the first item; and

displaying, by the one or more processors, the alert on a display device for viewing by one or more persons.

12 . The method of claim 11 , wherein applying the first weight to the first probability is based on the first identity having the characteristic.

13 . The method of claim 12 , wherein applying the second weight to the second probability is based on the second identity lacking the characteristic.

14 . The method of claim 11 , wherein assigning the first identity to the first item is in response to determining that (i) the first weighted probability exceeds the threshold and (ii) the second weighted probability falls below the threshold.

15 . The method of claim 11 , further comprising:

determining, based on the first image of the first item, a third probability for a third identity of the first item;

applying a third weight to the third probability based on the characteristic to produce a third weighted probability; and

in response to determining that both the first weighted probability and the third weighted probability exceed the threshold, prompting the user to select the first identity or the third identity, wherein assigning the first identity as the predicted identity of the first item is further based on the user selecting the first identity rather than the third identity.

16 . The method of claim 15 , wherein prompting the user comprises presenting, to the user, a second image for the first identity and a third image for the third identity.

17 . The method of claim 16 , wherein the first image and the third image are presented in an order based on the first weighted probability and the third weighted probability.

18 . The method of claim 15 , wherein prompting the user further comprises refraining from presenting, to the user, a fourth image for the second identity in response to determining that the second weighted probability falls below the threshold.

19 . The method of claim 11 , wherein assigning the first identity as the predicted identity of the first item is further based on at least one of a weight of the first item, an inventory count, or an expiration date.

20 . A non-transitory computer readable medium storing instructions that, when executed by a processor, cause the processor to:

obtain a scanned identity of an item based, at least in part, on a scanner scanning an identifier on the item;

capture, by a camera, an image of the item as the identifier is being scanned;

extract information about the item from the image;

determine, based on the information extracted from the image, a probability for an identity of the item;

determine, based on a shopping history of a user purchasing the item, a weight;

apply the weight to the probability to produce a weighted probability;

assign the identity as a predicted identity of the item based on the weighted probability;

determine a mismatch between the scanned identity of the item and the predicted identity of the item;

generate an alert indicating the mismatch between the scanned identity of the item and the predicted identity of the item; and

display the alert on a display device for viewing by one or more persons.