Abstract: Two models are first trained and then test images are applied to the two trained models in an effort to detect signature forgeries. The first model is trained with pairs of signature images and the resultant trained model is capable of detecting blind forgeries. The second model is trained with triplets of signature images and is capable of detecting skilled signature forgeries. After the two models are trained, test images are applied to the models and determinations are made as to whether a blind or skilled forgery is present.

Abstract: This application relates to systems and methods for automatically detecting and correcting image quality based on a set of quality standards. In some examples, a plurality of quality parameters of an image are determined based on receiving an image. It is then determined that at least one of the plurality of quality parameters is below a predetermined threshold. The predetermined threshold may be based on a required quality standard for images. In response to determining that the at least one of the plurality of quality parameters is below the predetermined threshold, feature of the image is adjusted such that the at least one of the plurality of quality parameters is at or above the predetermined threshold.

Abstract: Two models are first trained and then test images are applied to the two trained models in an effort to detect signature forgeries. The first model is trained with pairs of signature images and the resultant trained model is capable of detecting blind forgeries. The second model is trained with triplets of signature images and is capable of detecting skilled signature forgeries. After the two models are trained, test images are applied to the models and determinations are made as to whether a blind or skilled forgery is present.

Abstract: An automated planogram anomaly detection solution rapidly and reliably identifies mismatches between planograms and actual item placement. Examples receive a real time (RT) image of a shelf unit corresponding to at least a first portion of a planogram; detect, within the RT image, item boundaries for a plurality of items on the shelf unit and tag boundaries for a plurality of tags associated with the shelf unit; extract text from at least one tag; extract attributes from at least one item; map the extracted item attributes with the extracted tag text; detect, based at least on the map, a planogram anomaly; and based at least on detecting the planogram anomaly, generate a report identifying the planogram anomaly (e.g., a mismatch between a tag and an item). Some examples compare the RT image with a ground truth (GT) image to detect anomalies, for example empty space on the shelf unit.

Abstract: In various examples, a system receives image data characterizing an image of an item. Additionally, the system implements a first set of operations and a second set of operations. In some examples, the first set of operations includes performing a structural similarity analysis of the item, based on the image data, and determining a structural similarity score based on the structural similarity analysis of the item. In other examples, the second set of operations includes generating a plurality of derivative images by applying a plurality of distortions to the image of the item, extracting one or more features based at least on the plurality of derivative images, and determining the quality of the image based at least on the extracted one or more features and the structural similarity score.

Abstract: In various examples, a system receives image data characterizing an image of an item. Additionally, the system implements a first set of operations and a second set of operations. In some examples, the first set of operations includes performing a structural similarity analysis of the item, based on the image data, and determining a structural similarity score based on the structural similarity analysis of the item. In other examples, the second set of operations includes generating a plurality of derivative images by applying a plurality of distortions to the image of the item, extracting one or more features based at least on the plurality of derivative images, and determining the quality of the image based at least on the extracted one or more features and the structural similarity score.

Abstract: A systems including one or more processors and one or more non-transitory computer readable media storing computing instructions that, when executed on the one or more processors, perform: receiving a plurality of images from one or more devices, the images corresponding to a store shelf of a store; combining the plurality of images to generate a shelf image corresponding to the store shelf; encoding the shelf image into a first processing format; processing the shelf image in the first processing format with a neural network using pre-trained weights; determining positions in the shelf image that correspond to an out-of-stock detection based on outputs from the neural network; and generating a report for the out-of-stock detection, the report including an indication of coordinates of the out-of-stock detection and an item of the store that corresponds to the coordinates. Other embodiments are described.

Abstract: This application relates to automated processes for determining item placement compliance within retail locations. For example, a computing device may obtain an image of a fixture within a store. The image may be captured by a camera with a field of view directed at the fixture. The computing device may apply a segmentation process to the image to determine a portion of the image. Further, the computing device may determine a correlation between the portion of the image and each of a plurality of item image templates. Each item image template may include an image of an item the retail location sells in the retail location. The computing device may determine, based on the correlations, one of the plurality of item image templates and its corresponding item. The computing device may then determine whether the item should be located at the fixture based on a planogram.

Abstract: Two models are first trained and then test images are applied to the two trained models in an effort to detect signature forgeries. The first model is trained with pairs of signature images and the resultant trained model is capable of detecting blind forgeries. The second model is trained with triplets of signature images and is capable of detecting skilled signature forgeries. After the two models are trained, test images are applied to the models and determinations are made as to whether a blind or skilled forgery is present.

Abstract: This application relates to systems and methods for automatically detecting and correcting image quality based on a set of quality standards. In some examples, a plurality of quality parameters of an image are determined based on receiving an image. It is then determined that at least one of the plurality of quality parameters is below a predetermined threshold. The predetermined threshold may be based on a required quality standard for images. In response to determining that the at least one of the plurality of quality parameters is below the predetermined threshold, feature of the image is adjusted such that the at least one of the plurality of quality parameters is at or above the predetermined threshold.

Abstract: In various examples, a system receives image data characterizing an image of an item. Additionally, the system implements a first set of operations and a second set of operations. In some examples, the first set of operations includes performing a structural similarity analysis of the item, based on the image data, and determining a structural similarity score based on the structural similarity analysis of the item. In other examples, the second set of operations includes generating a plurality of derivative images by applying a plurality of distortions to the image of the item, extracting one or more features based at least on the plurality of derivative images, and determining the quality of the image based at least on the extracted one or more features and the structural similarity score.

Abstract: In some embodiments, apparatuses and methods are provided herein useful to automatically determining a discount for an item. In some embodiments, a system comprises an item database including information about a plurality of items, a control circuit configured to determine substitute items, calculate a score indicative of how similar a substitute item is to an item, determine a group of substitute items including items for which the score is above a threshold, determine a list of negotiable items, generate a user interface including the list of negotiable items, receive selection of one of the items from the list of negotiable items, determine the discount for the one of the items, and update the user interface to present the discount for the one of the items, a display device configured to present the user interface, and the user input device configured to receive the selection of the one of the items.

Abstract: Systems, methods, and computer-readable storage media for cataloguing and assessing images. This is performed by a system which receives images of an item, and identifying, within each image, the item. The system performs a structural similarity analysis of the item and for each image applies a plurality of distortions, such that for each image in the images multiple distorted images are generated. The system identifies within the distorted images at least one feature and applies a regression model to the images using the at least one feature and the structural similarity score.

Abstract: Systems, apparatuses, and methods are provided herein for automated food ingredient analysis. A method comprises retrieving a ingredients list associated with a food product, parsing the ingredients list to an ordered list of a plurality of ingredients, matching an ingredient of the plurality of ingredients to a reference ingredient to determine a nutrition content associated with the ingredient, assigning ingredient share ranking to the ingredient based on its position on the ingredients list, retrieving nutrition label information associated with the food product, and determining estimated ingredient shares of one or more of the plurality of ingredients in the food product based on the nutrition content of the one or more of the plurality of ingredients, the ingredient share ranking s of the one or more of the plurality of ingredients, and the overall nutrition content of the food product.

Abstract: An automated planogram anomaly detection solution rapidly and reliably identifies mismatches between planograms and actual item placement. Examples receive a real time (RT) image of a shelf unit corresponding to at least a first portion of a planogram; detect, within the RT image, item boundaries for a plurality of items on the shelf unit and tag boundaries for a plurality of tags associated with the shelf unit; extract text from at least one tag; extract attributes from at least one item; map the extracted item attributes with the extracted tag text; detect, based at least on the map, a planogram anomaly; and based at least on detecting the planogram anomaly, generate a report identifying the planogram anomaly (e.g., a mismatch between a tag and an item). Some examples compare the RT image with a ground truth (GT) image to detect anomalies, for example empty space on the shelf unit.

Abstract: Examples provide a system for performing retail-based cost reverse engineering and cost comparison within item similarity clusters for cost negotiations associated with a plurality of items. A cost manager component creates item similarity clusters based on item descriptions and cost trends associated with items over time. The cost manager component identifies items for action recommendations based on quantity and margin for each item. The recommendations can include a recommended set of item suitable for negotiations, promotional activities and/or a set of substitute items recommended for replacement of a selected item in an item assortment.

Abstract: In some embodiments, apparatuses and methods are provided herein useful to automatically determining a discount for an item. In some embodiments, a system comprises an item database including information about a plurality of items, a control circuit configured to determine substitute items, calculate a score indicative of how similar a substitute item is to an item, determine a group of substitute items including items for which the score is above a threshold, determine a list of negotiable items, generate a user interface including the list of negotiable items, receive selection of one of the items from the list of negotiable items, determine the discount for the one of the items, and update the user interface to present the discount for the one of the items, a display device configured to present the user interface, and the user input device configured to receive the selection of the one of the items.

Abstract: Two models are first trained and then test images are applied to the two trained models in an effort to detect signature forgeries. The first model is trained with pairs of signature images and the resultant trained model is capable of detecting blind forgeries. The second model is trained with triplets of signature images and is capable of detecting skilled signature forgeries. After the two models are trained, test images are applied to the models and determinations are made as to whether a blind or skilled forgery is present.

Abstract: Systems and methods are disclosed that enable more accurately generating customized alerts with computer vision (CV) and machine learning (ML), based on features extracted from collected imagery, when the features have contextual or situational significance. For example, a CV platform monitoring container content levels may generate an alert that may include placing a shopping cart entry or providing a purchase recommendation. Accompanying ML capability can leverage the CV historical data to improve container content and level determinations, and also adjust a consumption pattern threshold used as a trigger.

Abstract: A plurality of items are automatically clusterized at a first location. Social media postings for a given person are monitored and assessed at a second location in order to detect a trigger state. Upon detecting the trigger state the two locations automatically communicate with one another to identify one or more of the clusters that is relevant to the social media content. Information regarding items that relate to the identified cluster can then be provided to the user.