Abstract: Disclosed is a method and apparatus for operating a first device. The first device obtains environment information in proximity to the first device and receives one or more communication messages from a second device and the message(s) includes relevance criteria, wherein the relevance criteria indicates one or more devices, one or more sets of device characteristics, one or more lanes, one or more intersections or areas, one or more pedestrian paths or bicycle paths, one or more signal characteristics from the second device or any combination thereof. The first device determines whether the one or more communication messages are relevant to the first device based on the relevance criteria and the environment information and performs an operation in response to the determination of whether the one or more communication messages are relevant.

Abstract: In some aspects, a user device may receive, from a vision sensor of a user device, first image data associated with a first set of images. The user device may determine, using an image processing model, that a set of images depict a particular type of eye activity of an eye of a user. The user device may cause, based at least in part on determining that the first set of images depict the particular type of eye activity, a power level of a camera of the user device to be reduced.

Abstract: Various aspects of the present disclosure generally relate to a sensor module. In some aspects, an image sensor module may include an array of photon sensors configured to output a first set of signals corresponding to a set of photon sensors of the array of photon sensors. The set of photon sensors may include a row of photon sensors, or a column of photon sensors, of the array of photon sensors. The image sensor module may include a plurality of data selector components configured to receive the first set of signals and output a second set of signals corresponding to a subset of the set of photon sensors.

Abstract: Certain aspects of the present disclosure provide techniques and apparatus for performing mathematical operations on processing units based on data in the modulo space. An example method includes receiving a binary-space input to process (e.g., using a neural network or other processing system). The binary-space input is converted into a modulo-space input based on a set of coprimes defined for executing operations in a modulo space. A modulo-space result is generated through one or more modulo-space multiply-and-accumulate (MAC) units based on the modulo-space input. The modulo-space result is converted into a binary-space result, and the binary-space result is output.

Abstract: Aspects described herein provide a method including: receiving input data at a machine learning model, comprising: a plurality of processing layers; a plurality of gate logics; a plurality of gates; and a fully connected layer; determining based on a plurality of gate parameters associated with the plurality of gate logics, a subset of the plurality of processing layers with which to process the input data; processing the input data with the subset of the plurality of processing layers and the fully connected layer to generate an inference; determining a prediction loss based on the inference and a training label associated with the input data; determining an energy loss based on the subset of the plurality of processing layers used to process the input data; and optimizing the machine learning model based on: the prediction loss; the energy loss; and a prior probability associated with the training label.

Abstract: Various aspects of the present disclosure generally relate to optical character detection. In some aspects, a user device may receive, from a vision sensor, a first image that is associated with a first optical character image. The user device may determine, using an image processing model, that the first image depicts the first optical character image. The user device may cause, based at least in part on determining that the first image depicts the first optical character image, a camera to capture a second image that is associated with a second optical character image. The user device may perform an action associated with the second image. Numerous other aspects are provided.

Abstract: Disclosed are systems and techniques for transferring device content using radio frequency (RF) sensing. For instance, a first wireless device can identify a first user of the first wireless device based on a first radio frequency (RF) signature associated with the first user. The first wireless media device can determine a disengagement of the first user from the first wireless media device. In response to the disengagement, content information associated with usage of the first wireless device by the first user can be captured.

Abstract: Methods, systems, and devices for multi-device object tracking and localization are described. A device may transmit a request message associated with a target object to a set of devices within a target area. The request message may include an image of the target object, a feature of the target object, or at least a portion of a trained model associated with the target object. Subsequently, the device may receive response messages from the set of devices based on the request message. The response messages may include a portion of a captured image including the target object, location information of the devices, a pose of the devices, or temporal information of the target object detected within the target area by the devices. In some examples, the device may determine positional information with respect to the target object based on the one or more response messages.

Abstract: At least one feature pertains to an image sensor having an array of pixels with a configuration that includes a first pixel type configured to detect red light, a second pixel type configured to detect green light, a third pixel type configured to detect blue light and a fourth pixel type configured to detect green light and an array of lenses overlaying the array of pixels that comprises a first lens type, a second lens type, a third lens type and a fourth lens type where the first lens type is configured to pass red light and refract blue light and green light, the second lens type is configured to pass green light and refract blue light and red light, the third lens type is configured to pass blue light and refract red light and green light, and the fourth lens type is configured to pass green light and refract blue light and red light.

Abstract: Methods, systems, and devices for configuring a machine learning network are described. A device, which may be otherwise known as user equipment (UE), may support ultra-low power sensor applications. More specifically, the device may support memory block allocation of a machine learning network based on performance levels associated with the applications. For example, the device may identify a performance level associated with an application on the device. The device may determine that the performance level satisfies a condition, and subsequently determine a memory block allocation of a machine learning network of the device based on the performance level satisfying the condition. The memory block allocation may correspond to one or more connections of the machine learning network. Based on the memory block allocation, the device may adjust a quantity of memory blocks available for the machine learning network and process the application.

Abstract: In some aspects, a device may detect that an object is contacting a display of the device. The device may illuminate, based at least in part on detecting that the object is contacting the display, a plurality of pixel regions of the display. The plurality of pixel regions may be illuminated sequentially. The device may obtain, using one or more photosensors, a set of data relating to light that is reflected from at least a portion of the object based at least in part on illuminating the plurality of pixel regions. The set of data may include respective data for each sequential illumination of the plurality of pixel regions. The device may generate an image of at least the portion of the object based at least in part on the set of data. Numerous other aspects are described.

Abstract: Certain aspects of the present disclosure provide techniques for kernel expansion. An input data tensor is received at a first layer in a neural network, and a first convolution is performed for a first kernel, where the first kernel has a size greater than a preferred size. Performing the first convolution comprises generating a plurality of intermediate tensors by performing a plurality of intermediate convolutions using a plurality of intermediate kernels with a size of the preferred size, and accumulating the plurality of intermediate tensors to generate an output tensor for the first convolution.

Abstract: In some aspects, a device may detect that an object is contacting a display of the device. The device may illuminate, based at least in part on detecting that the object is contacting the display, a plurality of pixel regions of the display. The plurality of pixel regions may be illuminated sequentially. The device may obtain, using one or more photosensors, a set of data relating to light that is reflected from at least a portion of the object based at least in part on illuminating the plurality of pixel regions. The set of data may include respective data for each sequential illumination of the plurality of pixel regions. The device may generate an image of at least the portion of the object based at least in part on the set of data. Numerous other aspects are described.

Abstract: Techniques described herein include utilizing a mobile device as a proxy receiver for a vehicle in a V2X environment. In some embodiments, the mobile device may be configured to obtain vehicle and/or occupant metadata and store such data in memory at the mobile device. At a subsequent time, the mobile device may receive a V2X data message from a remote device (e.g., a V2X capable vehicle, another proxy device, a roadside unit, or any suitable device of the V2X environment that is different from the mobile device). The mobile device may process the data message based at least in part on the vehicle metadata and/or occupant metadata and perform one or more operations in response to processing the data message.

Abstract: Techniques described herein include utilizing a mobile device as a proxy receiver for a vehicle such that a driver of the vehicle can be presented driving assistance information based on messages received by the mobile device. The mobile device can determine that it is temporally located with the vehicle. If so, the mobile device can be configured to receive messages from other entities (e.g., vehicles, roadside units, traffic signals, and the like) in a vehicle-to-everything network. On reception, the mobile device may determine whether a received data message is relevant to the vehicle. In response to determining that the data message is, in fact, relevant to the vehicle, the mobile device may provide driving assistance information that is generated based at least in part on the information provided in the received data message.

Abstract: An example image capture device includes memory configured to store display content and image information received from a camera sensor, the camera sensor being configured to receive light through at least a portion of a display. The image capture device includes one or more processors coupled to the memory. The one or more processors are configured to determine a camera sensor blanking period. The one or more processors are configured to control the display to display content via one or more of a plurality of pixels in the at least a portion of the display during the camera sensor blanking period. The one or more processors are configured to control the display to not display content via the one or more of the plurality of pixels outside of the camera sensor blanking period.

Abstract: Techniques are provided for responding to congestion-sensitive preemptive data download requests in a V2X network. An example method for downloading data on a mobile device includes determining an estimated time of arrival at a station based at least in part on almanac information, generating a data download request based at least in part on the estimated time of arrival at the station, transmitting the data download request, and receiving one or more data packets from the station based on the data download request.

Abstract: Methods, systems, computer-readable media, and apparatuses for image capture are presented. An apparatus according to one aspect of the disclosure comprises a plurality of optical elements configured to direct light from an environment toward an image sensor. The apparatus further comprises one or more support structures coupled to the plurality of optical elements. According to this aspect, the one or more support structures are configured to support each of the plurality of optical elements at a relative location with respect to the image sensor. According to this aspect, each of the plurality of optical elements is configured to receive light from the environment based on a different field of view, as received light, and direct the received light toward the image sensor.

Abstract: Certain aspects of the present disclosure provide a method for compressing an activation function, comprising: determining a plurality of difference values based on a difference between a target activation function and a reference activation function over a range of input values; determining a difference function based on the plurality of difference values; and performing an activation on input data using the reference activation function and a difference value based on the difference function.

Abstract: A method for operating a low-bitwidth neural network includes converting a first activation to a non-negative value (e.g., absolute value). The first activation has a signed value. The sign of the activation is used to select a weight value. A product of the non-negative activation and the selected weight value is computed to determine a next activation. The next activation is quantized and supplied to a subsequent layer of the low-bitwidth neural network.