Abstract: A computer-implemented method and system for optimally retraining a supervised machine learning model based on newly received data. The method comprises receiving, from a requestor device, a new data set for updating a previously-trained model generated using a first training data set and tested using a first testing data set. Then, the new data set is checked for components having an association to both the first training data set and the first testing data set; and where such components are found, they are deleted. Once all of the components of the new data have been examined, remaining components of the new data set are assigned to one of the first training or testing data set in dependence upon a relationship connectivity therewith to form at least one of an updated testing and training data set for building the updated model.

Abstract: A computer-implemented is disclosed. The method includes: receiving, via a client device, a product search query; identifying a first product attribute based on query terms associated with the product search query; obtaining at least one search result matching the product search query; obtaining a set of product images associated with the at least one search result; for each of the product images, determining a respective level of confidence based on an estimation as to whether the product image is associated with the first product attribute; ranking the product images based on the levels of confidence; selecting at least a subset of the product images based on the ranking; and providing, via the client device, an indication of the at least one search result and the associated subset of product images.

Abstract: The present disclosure provides a search navigation system and method for an online store. The search navigation method includes obtaining attribute tags associated with available inventory associated with the online store, and detecting a buyer event associated with the online store. In response to the buyer event, a search navigation graph is generated for the online store based on a search navigation template graph. The search navigation graph comprises a hierarchical graph where each node in the search navigation graph represents a respective attribute tag, and the search navigation graph includes only nodes of the search navigation template graph representing attribute tags associated with the available inventory associated with the online store. A search navigation bar with selectable icons is displayed in a user interface. The selectable icons are a representation of a respective attribute tag represented by a respective node of the search navigation graph.

Abstract: Methods and systems for automatically ordering a set of images for consistent user interface display. The methods may include receiving a set of further images related to a first record, the first record referencing an ordered set of existing images, each of the existing images being assigned one or more respective image attributes. It may include assigning, using image analysis, one or more respective image attributes to each image in the set of further images and comparing image attributes assigned to the further images with image attributes assigned to the existing images to determine, for each of the further images, a corresponding one of the existing images. The further images are then ordered based on the determined corresponding ones of the existing images and the ordering of those existing images in the ordered set of existing images, and displayed in order in a user interface.

Abstract: A computer-implemented is disclosed. The method includes: obtaining video data and audio data for a live media stream; detecting a first product in at least one video frame of the live media stream, the first product being one of a first set of defined objects associated with the live media stream; identify one or more keywords in speech detected in the audio data, the one or more keywords being included in a second set of defined terms associated with the live media stream; determining a product variant of the first product based on the detected one or more keywords; and providing, for display via a client device associated with a viewer of the live media stream, an interactive user interface element associated with a first action in connection with the product variant.

Abstract: A computer application may aim to identify first and second “matching” objects. The matching method cannot necessarily be based on how visually similar the two objects are to each other because two matching objects might be different and/or be visually different. Moreover, the images of the objects to be matched might not necessarily have metadata to assist in the matching. In some embodiments, a machine learning model may be trained using a set of digital images, each including two or more matching objects. Triplet loss training may be used, and each triplet may include: an image of a first object extracted from a first image, an image of an object that is visually similar to an image of a second object extracted from the first image, and an image of a third object extracted from a different image.

Abstract: Systems and methods are disclosed for tire management that can detect various aspects of tires and determine tire status, from tires rotating on a moving vehicle.

Abstract: In order for the feature extractors to operate with sufficient accuracy, a high degree of training is required. In this situation, a neural network implementing the feature extractor may be trained by providing it with images having known correspondence. A 3D model of a city may be utilized in order to train a neural network for location detection. 3D models are sophisticated and allow manipulation of viewer perspective and ambient features such as day/night sky variations, weather variations, and occlusion placement. Various manipulations may be executed in order to generate vast numbers of image pairs having known correspondence despite having variations. These image pairs with known correspondence may be utilized to train the neural network to be able to generate feature maps from query images and identify correspondence between query image feature maps and reference feature maps. This training can be accomplished without requiring the capture of real images with known correspondence.

Abstract: In order for the feature extractors to operate with sufficient accuracy, a high degree of training is required. In this situation, a neural network implementing the feature extractor may be trained by providing it with images having known correspondence. A 3D model of a city may be utilized in order to train a neural network for location detection. 3D models are sophisticated and allow manipulation of viewer perspective and ambient features such as day/night sky variations, weather variations, and occlusion placement. Various manipulations may be executed in order to generate vast numbers of image pairs having known correspondence despite having variations. These image pairs with known correspondence may be utilized to train the neural network to be able to generate feature maps from query images and identify correspondence between query image feature maps and reference feature maps. This training can be accomplished without requiring the capture of real images with known correspondence.

Abstract: Systems and methods are disclosed for tire management that can detect various aspects of tires and determine tire status, from tires rotating on a moving vehicle.