Abstract: Image data is accessed. The image data includes frequency domain components. A subset of the frequency domain components is selected based on the relative importance of the frequency domain components. Only the subset of the frequency domain components is provided to an accelerator that executes a neural network to perform an artificial intelligence task using the subset of frequency domain components.

Abstract: A processing system comprises a processor and a memory. The memory stores instructions executable by the processor to receive a polarimetric image from a polarimetric camera sensor, and to identify a road surface in the received image based on a vehicle location, an orientation of the camera sensor, and a vehicle pose. The memory stores instructions, upon identifying, in the polarimetric image, polarized light reflections from the identified road surface based on a polarization direction and a polarization degree determined from the polarimetric image, to remove the identified polarized light reflections from the polarimetric image, thereby generating an updated polarimetric image including generating a de-mosaicked imaged based on the identified polarized light reflections, wherein the identified polarized light reflections are ignored at de-mosaicking, and to identify a road feature including a lane marking based on the updated polarimetric image.

Abstract: Methods, systems, and apparatus for controlling smart devices are described. In one aspect a method includes capturing, by a camera on a user device, a plurality of successive images for display in an application environment of an application executing on the user device, performing an object recognition process on the images, the object recognition process including determining that a plurality of images, each depicting a particular object, are required to perform object recognition on the particular object, and in response to the determination, generating a user interface element that indicates a camera operation to be performed, the camera option capturing two or more images, determining that a user, in response to the user interface element, has caused the indicated camera operation to be performed to capture the two or more images, and in response, determining whether a particular object is positively identified from the plurality of images.

Abstract: A crowd motion summarization method that provides a rich, real-time description of the crowd's characteristics from a video, such as, speed, orientation, count, spatial locations, and time. A feature tracking module receives each video frame and detects features (feature points) from the video frame. A crowd occupancy detection module receives the video frame and generates a binary crowd occupancy map having human pixel positions which indicate the human location versus non-human location, and generates a total human count of humans detected in the video frame. The feature tracking module generates feature tracking information for only those features contained in the human pixel positions which indicate the human location. In an example, the detected features are Kanade-Lucas-Tomasi (KLT) features. A feature-crowd matching module generates, using the feature tracking information and the total human count: crowd motion data. The method outputs the crowd motion data.

Abstract: Provided are a calibration method and a calibration apparatus for steering assist or autonomous driving of a vehicle in which lines on a road are detected from an image and coordinates of the lines in an image domain are transformed into coordinates in a world domain to update homography matrix that represents a transformation relationship between the image domain and the world domain, and controlling the vehicle traveling on the road based on the lines.

Abstract: Systems, methods and non-transitory computer readable media for allowing a user to switch between wearable items that have been paired or associated with an electronic device, such as a smartphone, are described. In one embodiment, the wearable items automatically detect a removal of a first wearable item from a user's body and an attachment of a second wearable item to the user's body. Messages from the wearable items are transmitted to the electronic device to allow the electronic device to switch the active wearable item from the first wearable item to the second wearable item. The switch can occur while the electronic device is in a locked state, and the electronic device can synchronize the second wearable item with data received from the first wearable item. Other embodiments are also described.

Abstract: Systems and methods for continuous monitoring of the behavior of animals, such as small rodents, are provided. Monitoring can include video, audio, and other sensor modalities. In one embodiment, the system can include cameras, arena design, environmental sensors, and ultrasonic sensors. The system uniquely provides a continuous long-term monitoring system suitable for mouse behavioral study. Further provided is a neural network based tracker configured for use with video data acquired by the monitoring system. 3 different neural network architectures have been tested to determine their performance on genetically diverse mice under varying environmental conditions. It has been observed that that an encoder-decoder segmentation neural network achieves high accuracy and speed with minimal training data.

Abstract: Systems, methods, tangible non-transitory computer-readable media, and devices for operating an autonomous vehicle are provided. For example, the disclosed technology can include receiving sensor data and map data. The sensor data can include information associated with an environment detected by sensors of a vehicle. The map data can include information associated with traffic signals in the environment. Further, an input representation can be generated based on the sensor data and the map data. The input representation can include regions of interest associated with images of the traffic signals. States of the traffic signals in the environment can be determined, based on the input representation and a machine-learned model. Traffic signal state data that includes a determinative state of the traffic signals can be generated based on the states of the traffic signals.

Abstract: There is provided with an information processing apparatus. An acquisition unit acquires data. A detection unit detects a determination target from the data. A training unit performs training of a determiner for performing category determination concerning the determination target using information of the determination target detected by the detection unit and intermediate data being generated during detection processing by the detection unit and being different from the information of the determination target.

Abstract: A communication system includes a transmitting module. The transmitting module is configured to send at least one preamble to a receiver module operating alternatively in an idle mode or a transmission mode during a synchronization time period (Tsync). The synchronization time period (Tsync) is greater than an idle time period of the idle mode of the receiving module.

Abstract: A passive incoherent millimeter-wave imaging system includes a receiver array including a plurality of receive modules configured to receive a scene signal reflected from a scene. The scene signal is reflected in response to a plurality of incoherent communication signals being reflected off the scene, and the plurality of incoherent communication signals are spatially and temporally incoherent at a point when reaching the scene.

Abstract: Technologies for monitoring beehives in an apiary includes one or more sensors communicatively coupled to a remote data analysis computing device, which is communicatively coupled to a client computing device. The sensors are configured to monitor beehives and/or the environment of the apiary and transmit the sensed data to the remote data analysis computing device. The remote data analysis computing device is configured to analyze the received sensor data, determine whether to perform an action as a function of a result of the analysis, and perform the function, as applicable. The client computing device is configured to interface with the remote data analysis computing device such that a user of the client computing device can view results of the data analysis and actions, as well as provide input and adjust settings administered by the remote data analysis computing device. A model for assessing beehive strength from an infrared image is also disclosed. Other embodiments are described herein.

Abstract: Systems, methods, and software are disclosed herein for associating images with content items. In an implementation, a computing device determines that an image captured of a scene includes a display of meeting content. The device identifies a content item comprising a source of the meeting content in the scene and may then create an association between an instance of the image and an instance of the content item. The device may, for example, create a link to the content item that can be embedded in or otherwise associated with the image file. The reverse may also be accomplished, whereby a link to the image may be created and embedded in or otherwise associated with the content item.

Abstract: Embodiments of the invention provide systems and methods of determining the authenticity of a document by creating a document fingerprint from known authentic documents and comparing the document to the document fingerprint. A plurality of authenticated documents may be compared to determine like characteristics. The documents containing like characteristics may be categorized based on the similarities while determining that the documents are of the same type, from a same company, and for a same user. Information from a data layer, user interface layer, and a device associated with the document may be obtained and a document fingerprint for the document created using the information. Upon receipt of a document submitted for authenticity, the document may be similarly categorized and compared to the document fingerprint for authenticity. A likelihood of authenticity and a fraud score may be determined and a rule, based on the fraud score and the likelihood, may be applied.

Abstract: Network devices (e.g., a modem, router, wireless user device, laptop, personal digital assistant or other similar wireless network devices) can be configured to monitor and detect a biological presence. In response to determining a biological presence (e.g., a human being or other similar being), a network device can alter parameters related to the generation of radio frequency (RF) energy in order to further ensure or guarantee safety from potential radiation as the number and power of network devices within a certain premises or vicinity increases.

Abstract: Described is a system for learning actions for image-based action recognition in an autonomous vehicle. The system separates a set of labeled action image data from a source domain into components. The components are mapped onto a set of action patterns, thereby creating a dictionary of action patterns. For each action in the set of labeled action data, a mapping is learned from the action pattern representing the action onto a class label for the action. The system then maps a set of new unlabeled target action image data onto a shared embedding feature space in which action patterns can be discriminated. For each target action in the set of new unlabeled target action image data, a class label for the target action is identified. Based on the identified class label, the autonomous vehicle is caused to perform a vehicle maneuver corresponding to the identified class label.

Abstract: An image processing apparatus comprises a specific object detection unit configured to detect a specific object from a moving image, a setting unit configured to set, in the moving image, based on a position of the specific object, a region of interest which is a region for performing an encoding process that produces a relatively higher image quality than in another region other than the region of interest, a moving object detection unit configured to detect a moving object from the moving image, and an encoding unit configured to perform an encoding process using, in the region of interest, an encoding parameter that can produce a relatively higher image quality than an encoding parameter used in the other region, when the moving object is detected in the region of interest.

Abstract: Example methods, apparatuses, systems, and articles of manufacture (e.g., physical storage media) to facilitate interference mitigation of aerial devices on a radio access network are disclosed. An example method includes receiving, by a processor from a base station of the radio access network, an indication of an interference event associated with the base station, and determining, by the processor, a set of unmanned aerial vehicles (UAVs) associated with the base station during flight of the set, where the interference event is at least partially attributed to each UAV of the set. The method further includes selecting, by the processor, a UAV among the set based on a respective traffic profile associated with each UAV of the set, and providing, by the processor, an instruction to the UAV to cause an adjustment of an operation of the UAV to reduce an interference impact of the UAV on the base station.

Abstract: A system for training transfer models that includes a memory and a processor coupling to each other. The memory stores instructions. The processor accesses the instructions to retrieve raw datasets from a first-type device and train a feature extraction model, a transfer feature and a preliminary classify model based on the raw datasets. The processor further transfers the feature extraction model, the transfer feature and the preliminary classify model to a host of a manufacturer so that the host directly applies the feature extraction model and the transfer feature, and retrains the preliminary classify model as an adapted classify model based on practical datasets.

Abstract: Disclosed is a vehicle. The vehicle comprises: storage in which a plurality of reference images and position information corresponding to respective plurality of reference images are stored; and a processor which acquires through a camera an image of the surrounding environment in which the vehicle is traveling, acquires a reference image corresponding to the current position of the vehicle among the plurality of reference images if the visibility of the surrounding environment image satisfies a predetermined condition, acquires an image in which the surrounding environment image and the reference image are merged, and controls the operation of the vehicle on the basis of the merged image.