Fuji Xerox Co., Ltd.

Fujifilm Business Innovation Corp.

Oath Inc.

Miteshkumar Patel

Andreas Girgensohn

Jacob Biehl

Chelhwon Kim

Donald G Kimber

Maani Ghaffari Jadidi

Patrick Chiu

Sven Kratz

Qiong Liu

David Ayman Shamma

Daniel Avrahami

Christian Holz

Matthew L Lee

Yusuke Yamaura

Xiaojing Zhang

Chidansh Amitkumar Bhatt

Senaka Buthpitiya

Jingwei Song

Raphael Guenot-Falque

Nikolas Hans-Friedrich Klug

Andrew Allan Port

Doga Buse Cavdir

A computer-implemented method, comprising detecting a first audio output in a first room, and detecting a portion of the first audio output in a second room, determining whether the portion of the first audio output in the second room meets a trigger requirement, and for the determining that the portion meets the trigger requirement, providing an action to reduce the portion of the first audio input in the second room.

System and method for dynamically adjusting settings of audio output devices to reduce noise in adjacent spaces

A computer-implemented method, comprising applying training images of an environment divided into zones to a neural network, and performing classification to label a test image based on a closest zone of the zones; extracting a feature from retrieved training images and pose information of the test image that match the closest zone; performing bundle adjustment on the extracted feature by triangulating map points for the closest zone to generate a reprojection error, and minimizing the reprojection error to determine an optimal pose of the test image; and for the optimal pose, providing an output indicative of a location or probability of a location of the test image at the optimal pose within the environment.

Fusing deep learning and geometric constraint for image-based localization

Example implementations described herein are directed to systems and methods for anomaly detection through using a segmentation process and an object detection process on images received through a camera system. The segmentation process and object detection process are then matched to detected additive anomalies (e.g., objects added to the environment) and subtractive anomalies (e.g., objects missing from the environment). Based on the type of anomaly detected as well as the underlying object, notifications can be dispatched to the user of the environment or the administrator of the system.

Understanding normality of an environment using semantic information from images

A computer-implemented method performed in a computerized system incorporating a central processing unit, a localization signal receiver, a plurality of sensors, separate and distinct from the localization signal receiver, and a memory, the computer-implemented method involving: using the central processing unit to initialize a plurality of particles based on an information on a map graph; using the central processing unit to repeatedly execute a particle filter loop, wherein the particle filter loop includes: using the central processing unit to perform a motion update of the plurality of particles; using the central processing unit to perform a measurement update of the plurality of particles; and using the central processing unit to perform a resampling of the plurality of particles based on particle importance weights and the map graph information. The location of the computerized system is subsequently determined based on the plurality of particles.

Systems and methods for utilizing graph based map information as priors for localization using particle filter

A computer implemented method is provided that includes embedding a received signal in a first modality, re-embedding the embedded received signal of the first modality into a signal of a second modality, and generating an output in the second modality, and based on the output, rendering a signal in the second modality that is configured to be sensed, wherein the embedding, re-embedding and generating applies a model that is trained by performing an adversarial learning operation associated with discriminating actual examples of the target distribution from the generated output, and performing a metric learning operation associated with generating the output having perceptual distances.

Transmodal translation of feature vectors to audio for assistive devices

Systems and methods of the present disclosure can involve several work spaces wherein each of the work spaces are associated with a set of activities and wherein each of the work spaces are coupled to one or more radio frequency (RF) sensors. Through RF sensor data detected from interactions with the work surface, example implementations described herein can determine which activity from the set of activities is being conducted through the application of a recognition algorithm that is generated from a machine learning algorithm.

User activity recognition through work surfaces using radio-frequency sensors

A computer-implemented method of localization for an indoor environment is provided, including receiving, in real-time, a dynamic query from a first source, and static inputs from a second source; extracting features of the static inputs by applying a metric learning convolutional neural network (CNN), and aggregating the extracted features of the static inputs to generate a feature transformation; and iteratively extracting features of the dynamic query on a deep CNN as an embedding network and fusing the feature transformation into the deep CNN, and applying a triplet loss function to optimize the embedding network and provide a localization result.

Indoor localization using real-time context fusion of visual information from static and dynamic cameras

System is built using the particle filter (PF) framework which utilizes data from round trip time (RTT) ranging or BLE signal strength to perform both (1) sensing the environment by using the scan data during measurement phase and (2) detect motion state of the user which it utilized during the motion update phase of PF. To detect the motion state of the user the temporal difference of the received ranging scans is used over which is utilized within the PF framework. The primary advantage of the described technique is that the velocity parameter of the PF can dynamically be updated based on the motion state as estimated using ranging scans.

Radiofrequency based virtual motion model for localization using particle filter

A probabilistic motion model to calculate the motion parameters of a user's hand held device using a noisy low cost Inertial Measurement Unit (IMU) sensor. Also described is a novel technique to reduce the bias noise present in the aforesaid IMU sensor signal, which results in a better performance of the motion model. The system utilizes a Particle Filter (PF) loop, which fuses radio signal data accumulated from Bluetooth Low Energy (BLE) beacons using BLE receiver with the signal from IMU sensor to perform localization and tracking. The Particle Filter loop operates based on a Sequential Monte Carlo technique well known to persons of ordinary skill in the art. The described approach provides a solution for both the noise problem in the IMU sensor and a motion model utilized in the Particle Filter loop, which provides better performance despite the noisy IMU sensor.

System and method for calibration-lessly compensating bias of sensors for localization and tracking

A novel solution to generate an optimal spatial placement map of Radio Frequency (RF) beacons, such as Bluetooth Low Energy (BLE) beacons, which can be used for indoor localization. A described embodiment solves for both generating optimal number of beacons required in a given environment and optimizing their location. The solution also incorporates behavior of RF signal due to various environmental factors such as signal reflection and absorption due to static and dynamic obstacles. In one embodiment, the framework is built upon Genetic Algorithm (GA) for optimization of beacon placement. The experimental results achieved using one implementation of the described method reduced the number of beacons by 50% as compared to beacons placed using expert knowledge. Furthermore, the overall localization error increased by &#x2dc;0.4 m when the results using the RF/BLE map generated by the described system were compared with the RF/BLE map generated using expert knowledge.

System and method for optimizing placement and number of radio-frequency beacons to achieve better indoor localization

A system and method, which utilizes incremental smoothing and mapping (iSAM) algorithm and automatically builds a beacon location map using various sensor and environmental information. The aforesaid iSAM algorithm fuses received signal strength indicator (RSSI) values available from different beacons in the environment and the information provided by the IMU sensor. The aforesaid iSAM algorithm is capable of simultaneously generating beacon and landmark map and localize the mobile computing device in the environment without having any prior information about any beacon locations. To accommodate for noisy sensor data and achieve better convergence for the iSAM algorithm, the system uses a prior beacon location map, which contains location information of some of the BLE beacons located in the environment. These known beacon locations provide cleaner environmental information to the iSAM algorithm and hence improve the overall estimation of beacon locations, which were not available apriori.

System and method for automating beacon location map generation using sensor fusion and simultaneous localization and mapping

Disclosed are systems and methods for improving interactions with and between computers in an authentication system supported by or configured with authentication servers or platforms. The systems interact to identify access and retrieve data across platforms, which data can be used to improve the quality of results data used in processing interactions between or among processors in such systems. The disclosed anti-spoofing systems and methods provide improved functionality to facial recognition systems by enabling enhanced 'spoof' (or attempts to impersonate a user) detection while authenticating a user. The disclosed systems and method provide additional functionality to existing facial recognition systems that enables such systems to actually determine whether the image being captured and/or recorded is that of an actual person, as opposed to a non-human representation.

Growing community of inventors

San Jose, CA, United States of America

Miteshkumar Patel