Abstract: A computing device retrieves a stabilized weight and a weight-based item count for one or more items to be purchased. Responsive to retrieving the stabilized weight, the computing device retrieves an image-based item count and an expected weight based on the image-based item count for the one or more items to be purchased. The computing device selects between authorizing and blocking checkout of the one or more items based on the weight-based item count, the image-based item count, the stabilized weight, and the expected weight.

Abstract: Computer vision grouping recognition is provided by receiving training images that include unpackaged items; identifying, by a computer vision model, a candidate identities for unpackaged items in a given training image; receiving, from a human user, a selected identity for the unpackaged item as feedback for the candidate identity; constructing a confusion matrix tallying matches and mismatches between candidate identities and the selected identities as analyzed across the training images for each unpackaged item; identifying at least one product category that includes at least a first unpackaged item and a second unpackaged item that the confusion matrix indicates as being misidentified for each other by the computer vision model; and reconfiguring the computer vision model to identify the product category instead of the first unpackaged item or the second unpackaged item when analyzing a given image including one or more of the first unpackaged item or the second unpackaged item.

Abstract: Systems and methods of providing navigation information associated with a retail space are provided. In one exemplary embodiment, a method comprises, a network node operationally coupled to a set of optical sensor devices positioned throughout a retail space, the network node receiving a sequence of images from each optical sensor device. The method comprises, obtaining a current position of a certain subject in the retail space based on one or more sequence of images received. The method comprises, obtaining a current position of an obstacle in the retail space based on one or more sequence of images received. The method comprises, sending to a wireless device associated with the certain subject, navigation information associated with the current positions of the certain subject and the obstacle in the retail space.

Abstract: Computer vision grouping recognition is provided by receiving training images that include unpackaged items; identifying, by a computer vision model, a candidate identities for unpackaged items in a given training image; receiving, from a human user, a selected identity for the unpackaged item as feedback for the candidate identity; constructing a confusion matrix tallying matches and mismatches between candidate identities and the selected identities as analyzed across the training images for each unpackaged item; identifying at least one product category that includes at least a first unpackaged item and a second unpackaged item that the confusion matrix indicates as being misidentified for each other by the computer vision model; and reconfiguring the computer vision model to identify the product category instead of the first unpackaged item or the second unpackaged item when analyzing a given image including one or more of the first unpackaged item or the second unpackaged item.

Abstract: This disclosure describes an automated process for training an ML model used by a computer vision system in a point of sale (POS) system to recognize a new item. Instead of relying on a manual process performed by a data scientist, the automated process can use images of a new (i.e., unknown) item captured at one or more POS systems to then retrain the ML model to recognize the new item. That is, the images of the item are used to retrain the ML model and to test the accuracy of the updated ML model. If the updated ML model can confidently identify the new item, the updated ML model is then used by the computer vision system to identify items at the POS system.

Abstract: Techniques for training an item recognition machine learning (ML) model are disclosed. An image of a first item for purchase is received. The image is captured by a point of sale (POS) system. A purchaser selection of a second item for purchase is also received. The purchaser makes the selection at the POS system. It is determined that the first item for purchase matches the second item for purchase, and in response training data is generated for an image recognition ML model, based on the image and the determination that the first item for purchase matches the second item for purchase. The ML model is trained using the training data, and the trained ML model is configured to recognize items for purchase in a plurality of images captured by a plurality of POS systems.

Abstract: Computer vision grouping recognition is provided by receiving training images that include unpackaged items; identifying, by a computer vision model, a candidate identities for unpackaged items in a given training image; receiving, from a human user, a selected identity for the unpackaged item as feedback for the candidate identity; constructing a confusion matrix tallying matches and mismatches between candidate identities and the selected identities as analyzed across the training images for each unpackaged item; identifying at least one product category that includes at least a first unpackaged item and a second unpackaged item that the confusion matrix indicates as being misidentified for each other by the computer vision model; and reconfiguring the computer vision model to identify the product category instead of the first unpackaged item or the second unpackaged item when analyzing a given image including one or more of the first unpackaged item or the second unpackaged item.