Abstract: Certain aspects of the present disclosure provide techniques for concurrently performing inferences using a machine learning model and optimizing parameters used in executing the machine learning model. An example method generally includes receiving a request to perform inferences on a data set using the machine learning model and performance metric targets for performance of the inferences. At least a first inference is performed on the data set using the machine learning model to meet a latency specified for generation of the first inference from receipt of the request. While performing the at least the first inference, operational parameters resulting in inference performance approaching the performance metric targets are identified based on the machine learning model and operational properties of the computing device. The identified operational parameters are applied to performance of subsequent inferences using the machine learning model.

Abstract: Certain aspects of the present disclosure provide techniques for concurrently performing inferences using a machine learning model and optimizing parameters used in executing the machine learning model. An example method generally includes receiving a request to perform inferences on a data set using the machine learning model and performance metric targets for performance of the inferences. At least a first inference is performed on the data set using the machine learning model to meet a latency specified for generation of the first inference from receipt of the request. While performing the at least the first inference, operational parameters resulting in inference performance approaching the performance metric targets are identified based on the machine learning model and operational properties of the computing device. The identified operational parameters are applied to performance of subsequent inferences using the machine learning model.

Abstract: Certain aspects of the present disclosure provide techniques for concurrently performing inferences using a machine learning model and optimizing parameters used in executing the machine learning model. An example method generally includes receiving a request to perform inferences on a data set using the machine learning model and performance metric targets for performance of the inferences. At least a first inference is performed on the data set using the machine learning model to meet a latency specified for generation of the first inference from receipt of the request. While performing the at least the first inference, operational parameters resulting in inference performance approaching the performance metric targets are identified based on the machine learning model and operational properties of the computing device. The identified operational parameters are applied to performance of subsequent inferences using the machine learning model.

Abstract: Techniques for associating environmental information with fingerprint images for fingerprint enrollment and matching are presented. The techniques may include capturing, using a fingerprint sensor of a mobile device, one or more images of a fingerprint. The techniques may include analyzing the one or more images to obtain fingerprint information associated with the fingerprint. The techniques may include obtaining, via an orientation sensor of the mobile device, environmental information indicating an orientation of the mobile device associated with the capturing of the one or more images. The techniques may additionally include generating, using the fingerprint information, an enrollment template for the fingerprint. The techniques may include associating the enrollment template with the environmental information.

Abstract: Systems and methods for biometric authentication are disclosed. The invention may be used to authenticate an individual. At least two enrollment templates may be obtained at different times. Later, an inquiry template may be obtained and compared to each of the enrollment templates in separate and distinct comparison operations. The match scores arising from the comparison operations may be fused to provide a composite match score. The composite match score may be compared to an acceptance range. If the composite match score is within the acceptance range, the individual may be authenticated.

Abstract: Embodiments of apparatus, computer program product, and method for verifying fingerprint images are disclosed.

Abstract: Certain aspects relate to systems and techniques for generating high resolution iris templates and for detecting spoofs, enabling more reliable and secure iris authentication. Pairs of RGB and NIR images can be captured by the iris authentication system for use in iris authentication, for example using an NIR LED flash and a four-channel image sensor. Multiple images of the user's iris can be captured by the system in a relatively short period of time and can be fused together to generate a high resolution iris image that can contain more detail of the iris structure and unique pattern than each individual images. The “liveness” of the iris, referring to whether the iris is a real human iris or an iris imitation, can be assessed via a liveness ratio based on comparison of known iris and sclera reflectance properties at various wavelengths to determined sensor responses at those same wavelengths.

Abstract: Techniques for associating environmental information with fingerprint images for fingerprint enrollment and matching are presented. The techniques may include capturing, using a fingerprint sensor of a mobile device, one or more images of a fingerprint. The techniques may include analyzing the one or more images to obtain fingerprint information associated with the fingerprint. The techniques may include obtaining, via an orientation sensor of the mobile device, environmental information indicating an orientation of the mobile device associated with the capturing of the one or more images. The techniques may additionally include generating, using the fingerprint information, an enrollment template for the fingerprint. The techniques may include associating the enrollment template with the environmental information.

Abstract: Methods, systems, computer-readable media, and apparatuses for novel eye tracking methodologies are presented. Specifically, after an initial determination of a person's eyes within a field of view (FOV), methods of the present disclosures may track the person's eyes even with part of the face occluded, and may quickly re-acquire the eyes even if the person's eyes exit the FOV. Each eye may be tracked individually, at a faster rate of eye tracking due to the novel methodology, and successful eye tracking even at low image resolution and/or quality is possible. In some embodiments, the eye tracking methodology of the present disclosures includes a series of sub-tracker techniques, each performing different eye-tracking functions that, when combined, generate a highest-confidence location of where the eye has moved to in the next image frame.

Abstract: Embodiments of apparatus, computer program product, and method for verifying fingerprint images are disclosed.

Abstract: A method of receiving user input by a mobile platform includes capturing a sequence of images with a camera of the mobile platform. The sequence of images includes images of a user-guided object in proximity to a planar surface that is separate and external to the mobile platform. The mobile platform then tracks movement of the user-guided object about the planar surface by analyzing the sequence of images. Then the mobile platform recognizes the user input based on the tracked movement of the user-guided object.

Abstract: Systems and methods for biometric authentication are disclosed. The invention may be used to authenticate an individual. At least two enrollment templates may be obtained at different times. Later, an inquiry template may be obtained and compared to each of the enrollment templates in separate and distinct comparison operations. The match scores arising from the comparison operations may be fused to provide a composite match score. The composite match score may be compared to an acceptance range. If the composite match score is within the acceptance range, the individual may be authenticated.

Abstract: Certain aspects relate to systems and techniques for generating high resolution iris templates and for detecting spoofs, enabling more reliable and secure iris authentication. Pairs of RGB and NIR images can be captured by the iris authentication system for use in iris authentication, for example using an NIR LED flash and a four-channel image sensor. Multiple images of the user's iris can be captured by the system in a relatively short period of time and can be fused together to generate a high resolution iris image that can contain more detail of the iris structure and unique pattern than each individual images. The “liveness” of the iris, referring to whether the iris is a real human iris or an iris imitation, can be assessed via a liveness ratio based on comparison of known iris and sclera reflectance properties at various wavelengths to determined sensor responses at those same wavelengths.

Abstract: Certain aspects relate to systems and techniques for generating high resolution iris templates and for detecting spoofs, enabling more reliable and secure iris authentication. Pairs of RGB and NIR images can be captured by the iris authentication system for use in iris authentication, for example using an NIR LED flash and a four-channel image sensor. Multiple images of the user's iris can be captured by the system in a relatively short period of time and can be fused together to generate a high resolution iris image that can contain more detail of the iris structure and unique pattern than each individual images. The “liveness” of the iris, referring to whether the iris is a real human iris or an iris imitation, can be assessed via a liveness ratio based on comparison of known iris and sclera reflectance properties at various wavelengths to determined sensor responses at those same wavelengths.

Abstract: Methods, systems, computer-readable media, and apparatuses for novel eye tracking methodologies are presented. Specifically, after an initial determination of a person's eyes within a field of view (FOV), methods of the present disclosures may track the person's eyes even with part of the face occluded, and may quickly re-acquire the eyes even if the person's eyes exit the FOV. Each eye may be tracked individually, at a faster rate of eye tracking due to the novel methodology, and successful eye tracking even at low image resolution and/or quality is possible. In some embodiments, the eye tracking methodology of the present disclosures includes a series of sub-tracker techniques, each performing different eye-tracking functions that, when combined, generate a highest-confidence location of where the eye has moved to in the next image frame.

Abstract: This disclosure presents methods, systems, computer-readable media, and apparatuses for optically tracking the location of one or more objects. The techniques may involve accumulation of initial image data, establishment of a dataset library containing image features, and tracking using a plurality of modules or trackers, for example an optical flow module, decision forest module, and color tracking module. Tracking outputs from the optical flow, decision forest and/or color tracking modules are synthesized to provide a final tracking output. The dataset library may be updated in the process.

Abstract: Techniques described herein relate to mobile computing device technologies, such as systems, methods, apparatuses, and computer-readable media for tracking an object from a plurality of objects. In one aspect, the plurality of objects may be similar. Techniques discussed herein propose dynamically learning information associated with each of the objects and discriminating between objects based on their differentiating features. In one implementation, this may be done by maintaining a database associated with each object and updating the dynamic database transferred while the objects are tracked. The tracker uses algorithmic means for differentiating objects by focusing on the differences amongst the objects. For example, in one implementation, the method may weigh the differences between different fingers higher than their associated similarities to facilitate differentiating the fingers.

Abstract: Techniques described herein relate to mobile computing device technologies, such as systems, methods, apparatuses, and computer-readable media for tracking an object from a plurality of objects. In one aspect, the plurality of objects may be similar. Techniques discussed herein propose dynamically learning information associated with each of the objects and discriminating between objects based on their differentiating features. In one implementation, this may be done by maintaining a database associated with each object and updating the dynamic database transferred while the objects are tracked. The tracker uses algorithmic means for differentiating objects by focusing on the differences amongst the objects. For example, in one implementation, the method may weigh the differences between different fingers higher than their associated similarities to facilitate differentiating the fingers.

Abstract: A circuit includes a circuit identification storage module and a control module. The circuit identification storage module stores circuit identification information. The control module receives the circuit identification information and in response thereto selectively performs a secure boot procedure or a test boot procedure. The control circuit performs the secure boot procedure when the circuit identification information indicates that the circuit is a production circuit. The control circuit performs the test boot procedure when the circuit identification information indicates that the circuit is a test circuit. A related method is also disclosed.

Abstract: A digital rights management system includes an authentication module and a decryption module. If desired, the modules can be implemented in separate integrated circuits. The authentication module retrieves authentication information for protected content and powers down after the authentication information is retrieved. The decryption module decrypts the protected content based on the authentication information while the authentication module is powered down.