Systems and methods for self-organizing data collection based on a production environment parameter are disclosed. An example monitoring system for data collection in a production environment may include a data collector coupled to a plurality of input channels coupled to a plurality of sensors co-located on a component of the production environment and to a network infrastructure; a data storage to store collected data; a data acquisition circuit to interpret a plurality of detection values, each of the plurality of detection values corresponding to at least one of the input channels; an expert system to self-organize data collection, wherein the self-organizing is based on a production parameter of the production environment; and wherein the data collector is responsive to the self-organizing to change a collection of the data.

In one example, a method performed by a processing system in a telecommunications network includes acquiring live footage of an event, acquiring sensor data related to the event, wherein the sensor data is collected by a sensor positioned in a location at which the event occurs, extracting an analytical statistic related to a target participating in the event, wherein the extracting is based on content analysis of the live footage and the sensor data, filtering data relating to the target based on the analytical statistic to identify content of interest in the data, wherein the data comprises the live footage, the sensor data, and data relating to historical events that are similar to the event, and generating computer-generated content to present the content of interest, wherein when the computer-generated content is synchronized with the live footage on an immersive display, an augmented reality media is produced.

Methods and systems for improving the interactivity of media content. The methods and systems are particularly applicable to the e-learning space, which features unique problems in engaging with users, maintaining that engagement, and allowing users to alter media assets to their specific needs. To address these issues, as well as improving interactivity of media assets generally, the methods and systems described herein provide for annotation and truncation of media assets. More particularly, the methods and systems described herein provide features such as annotation guidance and video condensation.

Images are tagged with values in an image data hierarchy that is most subjective at its top level and least subjective at its bottom level, such as a hierarchy including style, type, and features for clothing. A user preference hierarchy is determined from user response to images that are tagged. Tagged images may be generated by processing them with machine learning models trained to determine values for images. Product records including images and other data are analyzed to generate attribute vectors that are encoded to generate product vectors. Products are clustered according to their product vectors. Images of products within a cluster are clustered according to composition and groups of images are selected from image clusters for soliciting feedback regarding user preference for products of a cluster. Feedback is used to train a user preference model to estimate user affinity for a product having a given product vector.

An apparatus includes a display and a processor. The processor displays a virtual shopping cart. The processor also receives information indicating that an algorithm determined that a physical item was selected by a person during a shopping session in a physical store, based on a set of inputs received from sensors located within the store. In response, the processor displays a virtual item, which includes a graphical representation of the physical item. The processor additionally displays a rack video captured during the shopping session by a rack camera located in the store. The rack camera is directed at a physical rack located in the store, which includes the physical item. In response to displaying the rack video, the processor receives information identifying the virtual item, where the rack video depicts that the person selected the physical item. The processor then stores the virtual item in the virtual shopping cart.

A computing system generates a training data set for training the prediction model to detect defects present in a target surface of a target specimen and training the prediction model to detect defects present in the target surface of the target specimen based on the training data set. The computing system generates the training data set by identifying a set of images for training the prediction model, the set of images comprising a first subset of images. A deep learning network generates a second subset of images for subsequent labelling based on the set of images comprising the first subset of images. The deep learning network generates a third subset of images for labelling based on the set of images comprising the first subset of images and the labeled second subset of images. The computing system continues the process until a threshold number of labeled images is generated.

An approach is provided for filtering imagery to train a feature detection model. The approach involves, for example, receiving a plurality of images. The plurality of images is classified as depicting a feature of interest. The approach also involves providing data for presenting a bulk arrangement of at least one subset of the plurality of images, wherein the bulk arrangement is based on a characteristic of the plurality of images (e.g., detection confidence, position, size, visual characteristic, etc. of the detected feature). The approach further involves initiating a filtering of the plurality of images based on the bulk arrangement and providing the filtered plurality of images as training data to train the feature detection model.

An image processing apparatus acquires first image data obtained by capturing an image of a subject, second image data obtained by capturing an image of a person capturing the subject and surroundings thereof, and third image data indicating an appearance of an image capture apparatus that captures the image of the subject. The apparatus reduces a reflection in the first image data using a learned machine learning model that uses the first image data, the second image data, and the third image data as input data.

A system and a method to train a neural network are disclosed. A first image is weakly and strongly augmented. The first image, the weakly and strongly augmented first images are input into a feature extractor to obtain augmented features. Each weakly augmented first image is input to a corresponding first expert head to determine a supervised loss for each weakly augmented first image. Each strongly augmented first image is input to a corresponding second expert head to determine a diversity loss for each strongly augmented first image. The feature extractor is trained to minimize the supervised loss on weakly augmented first images and to minimize a multi-expert consensus loss on strongly augmented first images. Each first expert head is trained to minimize the supervised loss for each weakly augmented first image, and each second expert head is trained to minimize the diversity loss for each strongly augmented first image.

Some aspects of the present disclosure relate to systems, methods, and computer-readable media for configuring a computer system to detect violation conditions in a target dataset. In one example implementation, a computer implemented method includes: receiving data associated with an electronic communication; labelling the received data; creating a machine learning model based on the received data; creating a lexicon, where the lexicon represents one or more terms or regular expressions; creating a scenario using the machine learning models and the lexicon, where the scenario represents a violation condition; and configuring a computer system to detect violation conditions in a target dataset using the scenario, where the target dataset represents electronic communications.