Eastman-kodak Company

Nec Corporation

Xerox Corporation

Wei Zhu

Rajiv Mehrotra

Robert P Loce

Haifeng Chen

Cristian Lumezanu

Dongjin Song

Yuncong Chen

Takehiko Mizoguchi

Jiebo Luo

Edward B Gindele

Peter A Crean

Clara Cuciurean-Zapan

Robert T Gray

Ronald E Jodoin

George E Sotak

Methods and systems for training a neural network include collecting model exemplar information from edge devices, each model exemplar having been trained using information local to the respective edge devices. The collected model exemplar information is aggregated together using federated averaging. Global model exemplars are trained using federated constrained clustering. The trained global exemplars are transmitted to respective edge devices.

Federated learning for anomaly detection

Edge-side federated learning for anomaly detection

A system and method for determining image similarity. The method includes the steps of automatically providing perceptually significant features of main subject or background of a first image; automatically providing perceptually significant features of main subject or background of a second image; automatically comparing the perceptually significant features of the main subject or the background of the first image to the main subject or the background of the second image; and providing an output in response thereto. In the illustrative implementation, the features are provided by a number of belief levels, where the number of belief levels are preferably greater than two. The perceptually significant features include color, texture and/or shape. In the preferred embodiment, the main subject is indicated by a continuously valued belief map. The belief values of the main subject are determined by segmenting the image into regions of homogenous color and texture, computing at least one structure feature and at least one semantic feature for each region, and computing a belief value for all the pixels in the region using a Bayes net to combine the features. In an illustrative application, the inventive method is implemented in an image retrieval system. In this implementation, the inventive method automatically stores perceptually significant features of the main subject or background of a plurality of first images in a database to facilitate retrieval of a target image in response to an input or query image. Features corresponding to each of the plurality of stored images are automatically sequentially compared to similar features of the query image. Consequently, the present invention provides an automatic system and method for controlling the feature extraction, representation, and feature-based similarity retrieval strategies of a content-based image archival and retrieval system based on an analysis of main subject and background derived from a continuously valued main subject belief map.

System and method for determining image similarity

A method of analyzing one or more images of a user to determine the likelihood of user interest in materials that can be sent for display to the user includes selecting one or more images by a user; automatically analyzing the one or more user images to determine the likelihood that materials in a set will be of interest to the user; and selecting one or more items of materials based on their likelihood of interest to the user.

Analyzing images to determine if one or more sets of materials correspond to the analyzed images

A method for sharpening the corners of digital image data to overcome corner rounding when displayed or printed. The method comprises stepping a window across the image data and comparing that windowed data to a template. Upon determining a match, pixel data is either substituted into the image bitmap data at that location, or a signal is generated which will appropriately modify the writing of that image data. The data substitution or generated signal will achieve a localized clustering or &ldquo;ear&rdquo; of toner/ink, or in the alternative for inside corners the localized clustered absence of toner/ink. The result of this clustering is a sharpening of the corners so that they image as intended by the bitmap.

Image processing method for sharpening corners of text and line art

A method for retrieving digital patterns similar to a query pattern operates by generating representations of each digital pattern in terms of one or more color invariant properties; storing and organizing the patterns and their representations in a database; and searching the database by reference to the representations to identify and retrieve patterns similar to the query pattern. A preferred color invariant property is an edge property of the digital patterns. A web-based pattern registration and searching architecture enables on-line pattern search and selection utilizing a variety of pattern representations and similarity techniques, including the edge property based system described above.

Method and system for search and retrieval of similar patterns

A user preference for a desired image is determined by using one or more examples and counterexamples of a desired image in order to define the user preference. A relative preference of a user for one or more image components or one or more depictive features is automatically extracted from the examples and counterexamples of the desired image. Then, a user subjective definition of a desired image is formulated using the relative preferences for either the image components or the depictive features.

Method and computer program product for subjective image content similarity-based retrieval

The present invention is a method for antialiased rendering (AAR) of an image. First, tagging methods for identifying antialiased pixels were described, the methods being employed separately or in concert with one another. After identifying AAR pixels, a fill order is determined and applied using a fill order algorithm that preferably employs a pixel window that is a subset of the larger window employed for pixel identification.

Growing community of inventors

Rochester, NY, United States of America

Wei Zhu