Abstract: A method for validating performance of a neural network trained using labeled training and validation data is provided.

Abstract: Provided is a method for classifying data in an electronic apparatus, including obtaining target data, obtaining first classification information by using a classifier set including a plurality of classifiers based on the target data, obtaining second classification information by using a neural network model based on the target data, comparing the first classification information and the second classification information, and verifying the classifier set based on a result of comparing the first classification information and the second classification information.

Abstract: Disclosed is a method of processing information in an electronic apparatus, the method including acquiring a neural network model, determining a reference format for conversion of the neural network model, and converting the neural network model to a model of the reference format, wherein the model converted into the reference format is executed in a neural processing unit (NPU).

Abstract: The method and system evaluates the effectiveness of a display location within a store based on a behavioral response analysis of shoppers in the vicinity. The effectiveness of a display location is measured by tracking shoppers in-store, extracting and processing shopper attributes, and extracting metrics based on the processed attributes. The metrics of one location is compared to the metrics of another location to determine overall effectiveness.

Abstract: A method for enabling an labeling capability for training and validation data at an edge device to support neural network transfer learning capability is provided. The method includes: inputting candidate data into a first neural network to filter the candidate data by selecting a subset of candidate data based on an output of the first neural network, performing a confidence upgrade check on the subset of candidate data by: (1) performing a data consistency check by generating augmented data from each candidate data from among the subset of candidate data, (2) inputting the subset of candidate data into a second neural network that is trained using data from an environment to determine a second confidence condition, and (3) performing a clustering on the subset of candidate data, and automatically labeling, as training data, the subset of candidate data in accordance with a confidence level label.

Abstract: The present invention is a method and system for evaluating relative effectiveness of content in a digital signage network, using a series of measurements to compare and evaluate different contents. The measurements are performed on a plurality of input images of the people captured by a plurality of means for capturing images in the vicinity of at least a digital signage. The present invention is a content rating system based on how the audience responds to the digital content in the intended context of digital media network in-store or in other public viewing areas. The present invention allows the measurement of the response to the content for different segments, further allowing for better planning of the content and optimizing the advertising dollars.

Abstract: The present invention is a method and system for capturing a dataset over time that represents the response of shoppers to a set of marketing and merchandising stimuli or “elements”. The response includes exposure metrics, engagement metrics and conversion metrics for different shopper segments. The shopper segments may be defined by demographics such as gender, age and ethnicity or by the type of trip, such as a quick trip or a fill-up trip. The system comprises a plurality of means for capturing images, such as cameras, covering the area of interest in the vicinity of the marketing or merchandising element. The method comprises automated and semi-automated analysis of the video to extract the shopper behavior and demographics data for computing the defined metrics. The captured data can be further combined with information such a promotions and advertisement outside the store to further enhance the applications of the data collected in-store from the invention.

Abstract: The present invention is a method and system for detecting shopping groups based on the dynamic relation between shoppers' trajectories. First, shopper trajectories are generated using video images or positioning devices. Then, group behavior features are extracted from a candidate shopping group trajectory pair. The group behavior features of a given pair of trajectories are typically the changes in positional differences and the average speed of these trajectories. From a model of shopping group behavior, a given pair of candidate shopping group trajectories is analyzed to determine the group score—the likelihood of the pair of shoppers indeed belonging to the same group. Lastly, the system utilizes graph segmentation framework to find clusters where trajectories belonging to each cluster have tight group scores with each other.

Abstract: A method for enabling an labeling capability for training and validation data at an edge device to support neural network transfer learning capability is provided. The method includes: inputting candidate data into a first neural network to filter the candidate data by selecting a subset of candidate data based on an output of the first neural network, performing a confidence upgrade check on the subset of candidate data by: (1) performing a data consistency check by generating augmented data from each candidate data from among the subset of candidate data, (2) inputting the subset of candidate data into a second neural network that is trained using data from an environment to determine a second confidence condition, and (3) performing a clustering on the subset of candidate data, and automatically labeling, as training data, the subset of candidate data in accordance with a confidence level label.

Abstract: A method for validating performance of a neural network trained using labeled training and validation data is provided.

Abstract: The present invention is a system and method for measuring effectiveness of a marketing campaign on digital signage on many different signage networks, by measuring the efficiency of the campaign at reaching targeted audience and the effectiveness of conveying the message. This invention provides a solution to the challenges created by wide variety of measurements and lack of accuracy. By using automated audience measurement, the current invention is able to collect, large, statistically significant data for analysis. Non-intrusive, computer based measurement also ensure that the data is free from any biases. The media content rating system will provide a quantitative measure of how many people did the campaign reach and what effect did it have. The data will be available at the aggregate level, at network level and down to the screen level.

Abstract: The present invention is a method and system for rating in-store media elements based on the measurement for behavior patterns and demographics of the audience in the vicinity of the media element, where a plurality of input images of the audience are captured by at least a means for capturing images in the vicinity of the media element. The input images are processed by automated video analytic algorithms in order to measure the behavior patterns and demographics of each person in the audience tied to the media element. The measurements for the behavior patterns and demographics of the audience from the target media elements are aggregated. A set of ratings for the media element is calculated based on the measurements. The measured media elements are sampled, and the audience measurement for the sample is extrapolated to other media elements for in-store locations or for a group of stores. The ratings are converted into a standardized output.

Abstract: The present invention is a method and system for automatically recognizing which products a shopper intends to find or purchase based on the shopper's trajectory in a retail aisle. First, the system detects and tracks the person to generate the trajectory of the shopper. Then some of the dynamic features are extracted from the shopper trajectory. The shopper trajectory features of a given trajectory are typically the positions, the motion orientations, and speeds at each point of the trajectory. A shopper behavior model is designed based on some of the primitive actions of shoppers. The last step of the method is to analyze a given shopper trajectory to estimate the shopper's intention. The step either utilizes decision rules based on the extracted shopper trajectory features, or utilizes a trained Hidden Markov Model, to estimate the progression of the primitive actions from the trajectory.

Abstract: The present invention is a system and method for performing ethnicity classification based on the facial images of people, using multi-category decomposition architecture of classifiers, which include a set of predefined auxiliary classifiers that are specialized to auxiliary features of the facial images. In the multi-category decomposition architecture, which is a hybrid multi-classifier architecture specialized to ethnicity classification, the task of learning the concept of ethnicity against significant within-class variations, is handled by decomposing the set of facial images into auxiliary demographics classes; the ethnicity classification is performed by an array of classifiers where each classifier, called an auxiliary class machine, is specialized to the given auxiliary class. The facial image data is annotated to assign the age and gender labels as well as the ethnicity labels.

Abstract: The present invention provides a comprehensive method to design an automatic media audience measurement system that can estimate the site-wide audience of a media of interest (e.g., the site-wide viewership of a target display) based on the measurements of a subset of the actual audience sampled from a limited space in the site. This invention enables (1) the placement of sensors in optimal positions for the viewership data measurement and (2) the estimation of the site-wide viewership of the target display by performing the viewership extrapolation based on the sampled viewership data. The viewership extrapolation problem is formulated in a way that the time-varying crowd dynamics around the target display is an important decisive factor as well as the sampled viewership data at a given time in yielding the estimated site-wide viewership.

Abstract: The present invention is a system, method, and apparatus for determining the impact of crowding on retail performance based on a measurement for behavior patterns of people in a store area. The present invention captures a plurality of input images of the people by at least a means for capturing images, such as cameras, in the store area. In the captured plurality of input images, each person's shopping path is detected by a video analytics-based tracking algorithm. A subset of the people is identified as a crowd in the store area. In relation to the crowd, the behavior patterns of the target person are measured. After aggregating the measurements for the behavior patterns over a predefined window of time, the present invention can calculate a crowd index and a crowd impact index for the store area based on the measurements. A crowd index shows the level of crowd density in the store area caused by a crowd, including traffic count of the crowd in the store area.

Abstract: The present invention is a method and system for optimizing the observation and annotation of complex human behavior from video sources by automatically detecting predefined events based on the behavior of people in a first video stream from a first means for capturing images in a physical space, accessing a synchronized second video stream from a second means for capturing images that are positioned to observe the people more closely using the timestamps associated with the detected events from the first video stream, and enabling an annotator to annotate each of the events with more labels using a tool. The present invention captures a plurality of input images of the persons by a plurality of means for capturing images and processes the plurality of input images in order to detect the predefined events based on the behavior in an exemplary embodiment. The processes are based on a novel usage of a plurality of computer vision technologies to analyze the human behavior from the plurality of input images.

Abstract: The present invention is a method and system for producing a set of ratings for out-of-home media based on the measurement of behavior patterns and demographics of the people in a digital media network. The present invention captures a plurality of input images of the people in the vicinity of sampled out-of-home media in a digital media network by a plurality of means for capturing images, and tracks each person. The present invention processes the plurality of input images in order to analyze the behavior and demographics of the people. The present invention aggregates the measurements for the behavior patterns and demographics of the people, analyzes the data, and extracts characteristic information based on the estimated parameters from the aggregated measurements. Finally, the present invention calculates a set of ratings based on the characteristic information.

Abstract: The present invention is a method and system for measuring a set of shopper behavior metrics that represent the strength of a product category or a group of categories in the performance of a store area. A set of rating parameters are defined in order to provide a unified and standardized rating system. The rating system represents the effectiveness of the product category in a store area. The metrics are defined in a manner that is normalized so that they can be used across different types of product categories. The datasets are measured per category or group of categories over time to identify how the strength has varied over time, and to monitor trends in the category performance. The measured datasets are further analyzed based on various demographic groups and behavior segments. The analysis facilitates a better understanding of the strength of the category for different shopper segments, which in turn can be applied for developing better store area optimization strategies.

Abstract: The present invention is a method and system for selecting and storing videos by applying semantically-meaningful selection criteria to the track sequences of the trips made by people in an area covered by overlapping multiple cameras. The present invention captures video streams of the people in the area by multiple cameras and tracks the people in each of the video streams, producing track sequences in each video stream. The present invention determines a first set of video segments that contains the trip information of the people, and compacts each of the video streams by removing a second set of video segments that do not contain the trip information of the people from each of the video streams. The present invention selects video segments from the first set of video segments based on predefined selection criteria for the statistical behavior analysis.