NEC Corporation Of America

There is provided a computer implemented method of detecting data drift for changing a machine learning (ML) model, comprising: monitoring a target input dataset for being fed into the ML model, extracting a plurality of target candidate features from the target input dataset, applying a feature selection process for selecting a target subset of the plurality of target candidate features, accessing a historical input dataset previously determined as being suitable for being fed into the ML model, extracting a plurality of historical candidate features from the historical input dataset, applying the feature selection process for selecting a historical subset of the plurality of historical candidate features, computing a comparison between the target subset and the historical subset, and in response to the comparison meeting a requirement indicating a significant difference between the target subset and the historical subset, generating an indication for changing the ML model.

There is provided a computer implemented method of detecting an attempt to breach security of an iris recognition system by an artificial iris, comprising: analyzing at least a portion of a limbal ring depicted in an image of an iris of an individual captured by an imaging sensor at a wavelength range within at least one of near infrared (NIR) and short wave infrared (SWIR), and detecting likelihood of an artificial iris worn by the individual according to the analysis of the at least the portion of the limbal ring.

System and methods for enhancing the media consumption experience through generative AI-powered interactive companions. The system monitors media content in real-time, detects predefined cues, and generates contextually relevant and emotionally engaging reactions using a generative language model. These reactions are played and/or presented to the user, simulating an interactive conversation. The AI companion may adapt its responses based on user preferences, emotional states, and the context of the media content. By offering personalized interactions, emotional support, and/or contextual relevance, an immersive, engaging, and emotionally satisfying experience for users may be provided. The systems and methods may be used in entertainment, education, therapy, and social domains.

Disclosed herein is a method of improving computation of a 3 dimensional (3D) model of an object, comprising adjusting one or more illumination parameter of one or more light sources illuminating a target object having one or more high reflection surfaces, and operating one or more image sensors to capture a plurality of images depicting the target object from a plurality of different viewpoints while the object is illuminated by the one or more light sources. Wherein the plurality of images are used by one or more processors to compute a 3D model of the target object based on a plurality of features extracted from the plurality of images.

There is provided a system for computing a secure statistical classifier, comprising: at least one hardware processor executing a code for: accessing code instructions of an untrained statistical classifier, accessing a training dataset, accessing a plurality of cryptographic keys, creating a plurality of instances of the untrained statistical classifier, creating a plurality of trained sub-classifiers by training each of the plurality of instances of the untrained statistical classifier by iteratively adjusting adjustable classification parameters of the respective instance of the untrained statistical classifier according to a portion of the training data serving as input and a corresponding ground truth label, and at least one unique cryptographic key of the plurality of cryptographic keys, wherein the adjustable classification parameters of each trained sub-classifier have unique values computed according to corresponding at least one unique cryptographic key, and providing the statistical classifier, wherein the statistical classifier includes the plurality of trained sub-classifiers.

A method of detecting three dimensional (3D) changes in sites, comprising receiving a plurality of images of a site captured from a plurality of viewpoints by image sensor(s) mounted on dynamic vehicle(s) scanning the site, registering at least some of the images to each other, creating one or more merged representations of at least part of the site based on the at least some registered images, comparing between the merged representation(s) and one or more corresponding previous merged representations created based on a plurality of previous images captured during one or more previous scans of the site, detecting, based on the comparison, one or more 3D changes indicative of a depth change exceeding a certain threshold in one or more 3D features at the site, and generating a change indication indicative of the one or more detected 3D changes.

There is provided a computer implemented method of analyzing a sequence of images for detection of a change in face gesture, comprising: analyzing the sequence of time spaced images depicting a face of a subject to compute a baseline representation denoting a baseline state of the subject, for a respective image of the sequence of time spaced images depicting a face of a subject: extracting a plurality of features from the face of the subject of the respective image, computing a respective representation from the plurality of features, comparing the respective representation to the baseline representation, and detecting a change in face gesture of the subject when the comparison is above a threshold.

There is provided a computer implemented method of training a generative model, comprising: clustering training data elements each associated with an indication of a creation entity of creation entities, into clusters, each cluster including training data elements associated with one creation entity of the creation entities, generating training dataset by accessing training data elements from a sub-set of the clusters, each training dataset excluding at least one cluster of the clusters associated with at least creation entity, and training generative models on the training datasets, wherein each trained generative model of trained generative models is trained on training data elements that exclude at least one creation entity, wherein a target data element generated by a certain trained generative model in response to an input prompt excludes influence of the excluded at least one creation entity.

System and method for planar navigation for a moving platform utilizing computer vision techniques is disclosed herein. The method includes capturing with a monocular camera an image containing at least two distinct substantially vertical edges positioned around a waypoint, calculating the horizontal coordinate of the waypoint relative to the horizontal coordinates of the edges, and adjusting the moving platform's heading angle to maintain the said position of the navigation goal. The method includes tracking the pixel separation of the edges and terminating navigation of a segment responsive to the pixel separation exceeding a predetermined threshold.

A robot and method to lay cable from an unmanned aerial vehicle (“UAV”). The UAV adjusts its elevation and flight path to avoid obstacles on the ground and obstacles at elevation above the UAV and obstacles at elevation below the UAV.