Abstract: Methods, computer program products, and systems are presented. The methods include, for instance: generating a semantic network cell for a component of a semantic expression in a semantic network. The semantic network includes multiple semantic network cells. Each semantic network cell has attributes of a weight, an access count, and a latest time of access. A machine learning process reinforces the semantic network cell by access and deteriorates the semantic network cell over time based on semantic network cell weight rules, while the semantic network is servicing searches.

Abstract: The present invention provides, for instance, an apparatus for determining an abnormality in internal devices, wherein processing of determining the presence or absence of an abnormality in the internal devices is simplified. This ECU is an abnormality determining apparatus for determining the presence or absence of an abnormality in internal devices including: an image acquiring unit configured to acquire an image inside a vehicle; a luminance acquiring unit configured to acquire luminance of the image; and an LED configured to illuminate an occupant of the vehicle. The ECU is provided with an abnormality determining unit configured to determine, based on the luminance of the image, the presence or absence of an abnormality in the image acquiring unit and/or the LED.

Abstract: An apparatus for performing image analysis to identify human actions represented in an image, comprising: a joint-determination module configured to analyse an image depicting one or more people using a first computational neural network to determine a set of joint candidates for the one or more people depicted in the image; a pose-estimation module configured to derive pose estimates from the set of joint candidates that estimate a body configuration for the one or more people depicted in the image; and an action-identification module configured to analyse a region of interest within the image identified from the derived pose estimates using a second computational neural network to identify an action performed by a person depicted in the image.

Abstract: Disclosed is a method for detecting densely occluded fish based on a YOLOv5 network, belonging to the technical field of fish images. The method includes a data set establishment and processing part, a model training part and a model testing part; the data set establishment and processing part includes data collection of fish pictures, data labelling and data division of the fish pictures; and the data division is to divide data into a training set, a verification set and a test set.

Abstract: Some aspects of the subject matter disclosed herein include a system implemented on one or more data processing apparatuses. The system can include an interface configured to obtain, from one or more sensor subsystems, sensor data describing an environment of a vehicle, and to generate, using the sensor data, (i) one or more first neural network inputs representing sensor measurements for a particular object in the environment and (ii) a second neural network input representing sensor measurements for at least a portion of the environment that encompasses the particular object and additional portions of the environment that are not represented by the one or more first neural network inputs; and a convolutional neural network configured to process the second neural network input to generate an output, the output including a plurality of feature vectors that each correspond to a different one a plurality of regions of the environment.

Abstract: Apparatus and methods for enhanced 3D visualization includes receiving a plurality of images of an image scene from a plurality of image sensors. Depth information at locations of the image scene is received from a plurality of depth sensors. The depth information is combined with the plurality of images of the image scene using a machine learning model. A 3D representation of the image scene is generated based on the combined depth and image information.

Abstract: A vehicle controller includes a target position recognition unit configured to recognize a target position, a parking route generation unit configured to generate a parking route, and an automatic steering unit configured to automatically operate a steering device. The parking route generation unit generates a first route and a second route. The first route is a parking route to a temporary target position that is obtained by laterally moving the target position toward inside of the parking route in a turning direction. The second route is a parking route to the target position from a designated position in the middle of the first route. The automatic steering unit automatically operates the steering device to cause the subject vehicle to follow the first route until the subject vehicle reaches the designated position and to follow the second route from the designated position until the subject vehicle reaches the target position.

Abstract: There is provided an information processing apparatus including a reality determiner that determines whether an object detected on the basis of sensing data is a real object or an unreal object, and a controller that performs predetermined control on the basis of a result of the determination made by the reality determiner.

Abstract: In a method for facilitating creation of a map of a real-world, process control environment, locations of a mobile device are tracked as a user moves through a mapped environment. A camera of the mobile device captures images of the mapped environment as the user moves through the mapped environment, and the user indicates an intention to add a node to the map. One or more images of the captured images are provided to a machine learning (ML) model, and the ML model is trained to process images to recognize object types. The ML model may predict an object type corresponding to a specific object within a field of view of the camera. A display of the mobile device may then superimpose, on a real-world view presented to the user, an indication of the predicted object type to facilitate user designation of a descriptor for the new node.

Abstract: A method for acquiring a traffic state, an electronic device, a computer-readable storage medium, a roadside device and a cloud control platform are provided. An implementation of the method may include: acquiring monitoring data from at least one camera in a camera group, wherein the at least one camera is capable of shooting object moving states at different road sections or a same road section of a target intersection; determining, based on the monitoring data acquired from each camera in the at least one camera, a moving trajectory of a reference object recorded by the each camera in the at least one camera; then fusing moving trajectories of the reference object to obtain a completed trajectory; and finally predicting an indication state of a signal light on a corresponding road based on a travel direction of the corresponding road, the completed trajectory being located on the corresponding road.

Abstract: Example aspects of the present disclosure are directed to computing systems and methods for hand tracking using a machine-learned system for palm detection and key-point localization of hand landmarks. In particular, example aspects of the present disclosure are directed to a multi-model hand tracking system that performs both palm detection and hand landmark detection. Given a sequence of image frames, for example, the hand tracking system can detect one or more palms depicted in each image frame. For each palm detected within an image frame, the machine-learned system can determine a plurality of hand landmark positions of a hand associated with the palm. The system can perform key-point localization to determine precise three-dimensional coordinates for the hand landmark positions. In this manner, the machine-learned system can accurately track a hand depicted in the sequence of images using the precise three-dimensional coordinates for the hand landmark positions.

Abstract: Some embodiments are directed to, by a simple operation, appropriately displaying an image to which attention is paid. An information processing apparatus according to some embodiments is an information processing apparatus for performing display for measuring a width of a region in an image of an inspection target object, the information processing apparatus including an acquisition unit configured to acquire a bending position where a bend of a line approximating the region from the image satisfies a predetermined condition, and a display control unit configured to display, in an aligned manner in a predetermined direction, a plurality of partial images obtained by dividing the image based on the bending position.

Abstract: A head tracking system for use in a headrest, includes at least one optical circuit having a plurality of sensors. Each sensor arranged to detect light reflected from at least portion of an illuminated scene. The portion of the scene contains at least a part of a head to be tracked, and to provides an output signal corresponding to an amount of reflected light detected by the sensor. An acquisition circuit is arranged to receive a plurality of output signals from the plurality of sensors and to derive a frame from said output signals. A processing unit is arranged to receive the frame and to derive an indication of a change in position and/or orientation of the head by comparing the received frame with one or more frames obtained at different moments in time.

Abstract: A system for motion detection may include at least one storage medium that includes a set of instructions, and at least one processor in communication with the at least one storage medium. When executing the set of instructions, the at least one processor may be configured to cause the system to obtain data related to a video scene of a space from at least one video camera; detect an object in the video scene; classify the object as a human object or a non-human object; when the object is classified as a human object, track movements of the human object; and determine a posture of the human object in the video scene based on the movements of the human object.

Abstract: Methods, systems, and apparatus including computer programs encoded on a computer storage medium, for retrieving image search results using embedding neural network models. In one aspect, an image search query is received. A respective pair numeric embedding for each of a plurality of image-landing page pairs is determined. Each pair numeric embedding is a numeric representation in an embedding space. An image search query embedding neural network processes features of the image search query and generates a query numeric embedding. The query numeric embedding is a numeric representation of the image search query in the same embedding space. A subset of the image-landing page pairs having pair numeric embeddings that are closest to the query numeric embedding of the image search query in the embedding space are identified as first candidate image search results.

Abstract: According to one embodiment, a commodity registration device includes an optical reading unit with a reading range. The optical reading unit is configured to acquire an image of a commodity in the reading range. A wireless tag reading unit is configured to read a commodity code from a wireless tag of the commodity in the reading range. A processor determines whether the commodity is in the reading range of the optical reading unit and causes a notification unit to indicate the commodity has been registered if the wireless tag reading unit successfully reads the commodity code from the wireless tag while the commodity is determined to be in the reading range of the optical reading unit.

Abstract: Classification of cancer condition, in a plurality of different cancer conditions, for a species, is provided in which, for each training subject in a plurality of training subjects, there is obtained a cancer condition and a genotypic data construct including genotypic information for the respective training subject. Genotypic constructs are formatted into corresponding vector sets comprising one or more vectors. Vector sets are provided to a network architecture including a convolutional neural network path comprising at least a first convolutional layer associated with a first filter that comprise a first set of filter weights and a scorer. Scores, corresponding to the input of vector sets into the network architecture, are obtained from the scorer. Comparison of respective scores to the corresponding cancer condition of the corresponding training subjects is used to adjust the filter weights thereby training the network architecture to classify cancer condition.

Abstract: A medical image learning method of a medical image process apparatus includes preparing a plurality of body X-ray images for learning as an input of a learning data set, preparing internal diagnostic indicator information corresponding to each of the plurality of body X-ray images for learning as a label of the learning data set, and learning an artificial neural network model using the learning data set. The internal diagnostic indicator information includes information on a cardiovascular border. The information on the cardiovascular border includes information on at least one of an aortic knob, a pulmonary conus, a left atrial appendage, an upper right cardiac border, a lower right cardiac border, a lower left cardiac border, a descending aorta, a carina, an upper end point of a diaphragm, a right pulmonary artery, a posterior cardiac border, and an anterior spinal border.

Abstract: An image processing device according to one aspect of the present disclosure includes: at least one memory storing a set of instructions; and at least one processor configured to execute the set of instructions to: receive a visible image of a face; receive a near-infrared image of the face; adjust brightness of the visible image based on a frequency distribution of pixel values of the visible image and a frequency distribution of pixel values of the near-infrared image; specify a relative position at which the visible image is related to the near-infrared image; invert adjusted brightness of the visible image; detect a region of a pupil from a synthetic image obtained by adding up the visible image the brightness of which is inverted and the near-infrared image based on the relative position; and output information on the detected pupil.

Abstract: A prevention and safety management system utilizes a non-intrusive imaging sensor (e.g. surveillance cameras, smartphone cameras) and a computer vision system to record videos of workers not wearing sensors. The videos are analyzed using a deep machine learning algorithm to detect kinematic activities (set of predetermined body joint positions and angles) of the workers and recognizing various physical activities (walk/posture, lift, push, pull, reach, force, repetition, duration etc.). The measured kinematic variables are then parsed into metrics relevant to workplace ergonomics, such as number of repetitions, total distance travelled, range of motion, and the proportion of time in different posture categories. The information gathered by this system is fed into an ergonomic assessment system and is used to automatically populate exposure assessment tools and create risk assessments.

Abstract: An image processing device according to one aspect of the present disclosure includes: at least one memory storing a set of instructions; and at least one processor configured to execute the set of instructions to: receive a visible image of a face; receive a near-infrared image of the face; adjust brightness of the visible image based on a frequency distribution of pixel values of the visible image and a frequency distribution of pixel values of the near-infrared image; specify a relative position at which the visible image is related to the near-infrared image; invert adjusted brightness of the visible image; detect a region of a pupil from a synthetic image obtained by adding up the visible image the brightness of which is inverted and the near-infrared image based on the relative position; and output information on the detected pupil.

Abstract: Techniques related to training and implementing a bidirectional pairing architecture for object detection are discussed. Such techniques include generating a first enhanced feature map for each frame of a video sequence by processing the frames in a first direction, generating a second enhanced feature map for frame by processing the frames in a second direction opposite the first, and determining object detection information for each frame using the first and second enhanced feature map for the frame.

Abstract: A method, an apparatus, a computing device, and a computer-readable storage medium for monitoring use of target item are disclosed. The method includes obtaining a RoI of each of video frames, for each frame, determining whether the target item and a head of a monitored target object are both present in the RoI, if yes, determining whether a hand of the monitored target object is present in the RoI, if the hand not present, determining a face orientation of the monitored target object to determining whether the each frame is in a first video frame status, if the hand presents, determining a relative position relationship between the hand and the target item to determining whether the each frame is in the first video frame status, and based on a number of video frames continuously in the first video frame status, determining whether the monitored target object uses the target item.

Abstract: An issue tracking system for tracking software development tasks is described herein. The issue tracking system may be configured to receive new issue requests from a client device and associate the new issue requests with one or more clusters of previously stored issue records. The issue tracking system may also determine similarity between issues in a first cluster of stored issue records and issues in a second cluster that is associated with a different software development project. Based on a determination that the issue similarity exceeds a threshold, the user may be prompted with one or more recommendations for a subsequent issue request or issue request content.

Abstract: One embodiment provides techniques for automatically pre-labeling point cloud data with cuboid annotations. Point cloud data is processed using ML models to detect, associate, and localize objects therein, in order to generate cuboid tracks that each include a series of cuboid annotations associated with an object. An object detection model that detects objects and performs coarse localization is trained using a loss function that separately evaluates the distances between corners of predicted cuboids and corners of ground truth cuboids for position, size, and yaw. A refinement model that performs more accurate localization takes as input 2D projections of regions surrounding cuboid tracks predicted by the object detection model and the cuboid tracks, and outputs refined cuboid tracks. The refined cuboid tracks are filtered to a set of keyframes, with in-between frames being interpolated. The cuboid tracks can then be presented to a user for viewing and editing.

Abstract: A method of controlling a vehicle includes obtaining, by a camera, an image of a road ahead; recognizing, by a controller, a curvature of a front lane from the road image and obtaining position and speed information of another vehicle based on obstacle information detected by an obstacle detector; periodically storing, by a storage, driving speed information, yaw rate information, and steering angle information while driving; recognizing, by the controller, a curvature of a rear lane based on the driving speed information, yaw rate information, and steering angle information in response to determining that a lane change is necessary; determining, by the controller, a lane change possibility based on the curvature of the front lane, the curvature of the rear lane, and the position speed information of the other vehicle; and controlling, by the controller, at least one of steering, deceleration, and acceleration based on the lane change possibility.

Abstract: A biometric capture device is operative to adjust one or more environmental parameters to enhance a range (e.g., distance) at which a biometric may be captured from a subject. For example, a sample biometric capture device may be, include, or otherwise incorporate a retinal or iris scanner configured to capture an image of the retina or iris (e.g., a biometric) when the retina or iris is illuminated by infrared light. Generally, the amount of infrared light required to accurately image the retina or iris increases with the distance of the subject's retina or iris from the image capture device. The biometric capture device may capture a facial image using a first image sensor, identify a face in the facial image, capture an iris image using a second image sensor guided by the facial image, and identify a person using the iris image.

Abstract: A method for detecting the presence of on-body concealed objects includes receiving a visible-domain camera image for a scene, determining, using the visible-domain camera image, a region of interest where a subject is present, receiving an infrared-domain camera image and a millimeter-wave (mmwave) radar image that each cover the region of interest, determining emissivity information for the region of interest using the infrared-domain camera image, determining reflectivity information for the region of interest using the mmwave radar image and determining a concealed object classification for the subject based on the emissivity information and the reflectivity information. A corresponding system and computer program product for executing the above method are also disclosed herein.

Abstract: A computing system for decimating video data includes: a processor; a persistent storage system coupled to the processor; and memory storing instructions that, when executed by the processor, cause the processor to decimate a batch of frames of video data by: receiving the batch of frames of video data; mapping, by a feature extractor, the frames of the batch to corresponding feature vectors in a feature space, each of the feature vectors having a lower dimension than a corresponding one of the frames of the batch; selecting a set of dissimilar frames from the plurality of frames of video data based on dissimilarities between corresponding ones of the feature vectors; and storing the selected set of dissimilar frames in the persistent storage system, the size of the selected set of dissimilar frames being smaller than the number of frames in the batch of frames of video data.

Abstract: A Head-Mounted Display with camera sensors to perform chroma keying in a mixed reality context is presented. Low latency is achieved by embedding the processing in the HMD itself, specifically, format camera images, detect the selected color range and make a composite with the virtual content.

Abstract: In some aspects, the techniques described herein relate to systems, methods, and computer readable media for detecting movement in a scene. A first temporal pixel image is generated based on a first set of images of a scene over time, and a second temporal pixel image is generated based on a second set of images. One or more derived values are determined based on values of the temporal pixels in the first temporal pixel image, the second temporal pixel image, or both. Correspondence data is determined based on the first temporal pixel image and the second temporal pixel image indicative of a set of correspondences between image points of the first set of images and image points of the second set of images. An indication of whether there is a likelihood of motion in the scene is determined based on the one or more derived values and the correspondence data.

Abstract: A system includes a plurality of sensors distributed about a space. Each sensor generates sensor data associated with properties of one or more people in the space. A gateway device receives the sensor data from each of the plurality of sensors and provides the sensor data to an automated response subsystem over a network. The automated response subsystem receives the sensor data. A first feature value is determined based on the received sensor data. An event score is determined based on the first feature value. The event score is compared to predefined response criteria. Based on the comparison of the event score to the predefined response criteria, an actionable event is detected, and an automated response to initiate for the detected actionable event is identified. One or both of an alert and instructions are provided in order to initiate the automated response.

Abstract: A heterogeneous sensor fusion apparatus includes a point processor configured to detect a first object by processing a detection point input from a first sensor, an image processor configured to detect a second object by processing an image input from a second sensor, a point-matching unit configured to calculate a matching rate by matching the detection point with the image, and to determine whether the first object and the second object are identical based on the calculated matching rate, an association unit configured to generate track information by fusing information from the first sensor and the second sensor when the first object and the second object are identical, and an output unit configured to output the generated track information.

Abstract: An object detection system includes a first camera having a first field of view and a second camera having a second field of view that overlaps with the first field of view. The object detection system includes object detection logic configured to receive a first image from the first camera and a second image from the second camera and identify an object in both the first and second image. The object detection system further includes object mapping logic configured to identify a location of the object based on the first and second image. The object detection system also includes display logic configured to display the location of the object.

Abstract: Systems and methods are provided herein for performing one or more actions based on kinematic data. An operator management module may obtain video input data depicting a two-dimensional representation of a subject during performance of an activity. A set of points of the subject may be identified based at least in part on a machine-learning model. A 3D representation of the subject may be generated based at least in part on the points identified. Kinematic data related to the subject may be generated utilizing the 3D representation of the subject. Any suitable number of suggested actions may be performed based at least in part on the kinematic data.

Abstract: A detection device for detecting the presence of a substance of interest in a sample is described. The device can include a data store comprising executable instructions for at least one convolutional neural network, CNN, configured to process images: and a processor coupled to the data store and configured to execute the instructions to operate the at least one CNN. The detection device can be configured to: obtain spectrometry data, operate a first one of the CNNs to process the spectrometry data to obtain a first CNN output; apply a mask to the spectrometry data to obtain masked data; operate a second one of the CNNs to process the masked data to obtain a second CNN output; and determine if the substance of interest is present in the sample based on both the first CNN output and the second CNN output.

Abstract: Aspects of the present disclosure are directed to an artificial reality (XR) application system controlling applications in an artificial reality environment. In various cases, these controls include automatically suggesting XR applications by determining an XR context and identifying applications that match the XR context. These applications can be suggested to a user, who can authorize their execution, setting permissions for the application. In some cases, applications can be divided into components which can be progressively downloaded. By providing application suggestions relevant to the current context and progressively downloading application components, applications can appear ambient, rather than relying on users to constantly download, install, or activate applications. Permissions for applications may be revoked permanently or for certain situations—either through user permissions selections or automatically in response to determined user intents.

Abstract: An exemplary aspect is a location estimation system, including: a feature value extraction processing unit that performs feature value extraction processing for extracting a feature value from an input image; and a location estimation processing unit that compares the extracted feature value with map information and performs location estimation processing for estimating a self-location based on a result of the comparison. While the location estimation processing unit is performing the location estimation processing corresponding to an input image input at a predetermined timing, the feature value extraction processing unit performs the feature value extraction processing on an input image input at a timing after the predetermined timing.

Abstract: A computer-implemented method of signal processing comprises providing images. The method comprises for each respective one of at least a subset of the images: applying a weakly-supervised learnt function, the weakly-supervised learnt function outputting respective couples each including a respective localization and one or more respective confidence scores, each confidence score representing a probability of instantiation of a respective object category at the respective localization. The method further comprises determining, based on the output of the weakly-supervised learnt function, one or more respective annotations, each annotation including a respective localization and a respective label representing instantiation a respective object category at the respective localization. The method further comprises forming a dataset including pieces of data, each piece of data including a respective image of the subset and at least a part of the one or more annotations determined for the respective image.

Abstract: A dishwasher includes: a tub, a sump disposed in the tub, a rack disposed in the sump and receiving an object, a plurality of lighting devices disposed at the tub and configured to illuminate an upper end portion of the rack, a measuring camera disposed at the tub and capturing an image of the upper end portion of the rack and generating an object image, and a controller configured to perform a first lighting control on the lighting devices, acquiring a first object image generated based on a first image of the rack captured by the measuring camera, performing a second lighting control on the lighting devices, acquiring a second object image generated based on a second image of the upper end portion of the rack captured by the object measuring camera, and generating a three-dimensional shape image of the object based on the first and second object images.

Abstract: A monitoring system according to the present disclosure includes a cable (20) comprising an optical fiber, a reception unit (31) configured to receive an optical signal including a pattern corresponding to a state of a monitoring target (10) from at least one optical fiber included in the cable (20) and to detect the pattern from the received optical signal, and a control unit (32) configured to detect the state of the monitoring target (10) based on the pattern.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for tracking moving objects depicted in multiple images. One of the methods includes determining, for an image captured by a camera, a first bounding box that represents a first moving object depicted in the image, determining that the first bounding box and a second bounding box overlap in an overlap area, determining that the first moving object represented by the first bounding box was farther from the camera that captured the image than a second moving object represented by the second bounding box, generating a mask for the first bounding box based on the overlap area, and determining, using data from the image that is associated with the mask, that the first moving object matches an appearance of another moving object depicted in another image captured by the camera.

Abstract: A data enhancement expansion method, recognition method and system for deep learning, the data enhancement expansion method includes the following steps: collecting original video data of a target to be recognized, and extracting original images of several recognized targets from the original video data; extracting seed images of RoI outlines of the recognized targets from the original images; performing an image enhancement operation on the seed images of the RoI outlines of the recognized targets, and randomly extracting the seed images subjected the image enhancement operation for image aliasing enhancement to obtain several composite images; and generating a data set based on the original images of the several recognized targets and the several composite images. Original data materials are easy to obtain with extremely low cost and high authenticity, and can be really put to a deep learning network to achieve good recognition results.

Abstract: A neural network is configured to suppress an output of a mitigation node in a mitigation layer of the neural network. The neural network is pre-configured to recognize objects from inputs when operating using a processor and a memory. An actual input is sent to the neural network for object recognition, the actual input is an altered input. By suppressing the output of the mitigation node, the neural network is caused to avoid falsely recognizing an object from the actual input, where the altered input is configured to cause the neural network to falsely recognize the object from the actual input.

Abstract: An example system includes a processor to receive an image with corresponding acquisition information. The processor is to classify the image using the corresponding acquisition information via a deep neural network including integrated acquisition information.

Abstract: Example aspects include a method, apparatus and computer-readable medium of enforcing dynamic social distancing rules, comprising monitoring video frames of one or more video feeds, classifying a plurality of persons in the video frames, calculating respective distances between each person of the plurality of persons and remaining persons of the plurality of persons, detecting mask usage of the plurality of persons, measuring a dwell time of the plurality of persons, determining whether the plurality of persons are in violation of the dynamic social distancing rules, and generating an alert. The mask usage indicating whether each person of the plurality of persons is wearing a mask. The dwell time indicating a duration of time that each person of the plurality of persons has been within a proximity threshold to the remaining persons of the plurality of persons.

Abstract: Techniques for determining map data for localizing a vehicle in an environment are discussed herein. A vehicle can capture sensor data such as image data and can detect objects in the sensor data. Detected objects can include semantic object such as traffic lights, lane lines, street signs, poles, and the like. Rays based on a pose of the sensor and the detected objects can be determined. Intersections between the rays can be determined. The intersections can be utilized to determine candidate landmarks in the environment. The candidate landmarks can be utilized to determine landmarks and to eliminate other candidate landmarks. The map data can be determined based on the three-dimensional locations and utilized to localize vehicles.

Abstract: A vehicle surveillance assembly includes an inside video camera mounted within a cabin of a vehicle to record video footage of an interior of the cabin. A plurality of outside video cameras is each mounted to an exterior of the vehicle to record video footage of the environment surrounding the vehicle. A tracking unit is wearable on the driver of the vehicle and the tracking unit broadcasts a tracking signal. A control unit is positioned within the vehicle and the control unit is in remote communication with the tracking unit, the inside video camera and each of the plurality of outside video cameras. The control unit actuates the inside video camera when the control unit determines the tracking unit is inside of the vehicle. Conversely, the control unit actuates each of the plurality of outside video cameras when the control unit determines the tracking unit is outside of the vehicle.

Abstract: Certain aspects provide a method of generating a physical activity model, including: receiving, via a motion capture device, motion data corresponding to a plurality of key states associated with a physical activity sequence; for each respective key state in the plurality of key states: determining a plurality of joint positions associated with the respective key state; determining a plurality of body segment positions associated with the respective key state based on the plurality of joint positions; determining a plurality of inter-state differentiation variables for the respective key state; determining one or more state characteristic metrics for the respective key state; and determining a classifier for the respective key state based on the one or more state characteristic metrics; and defining a physical activity model based on the one or more state characteristic metrics and the classifier associated with each key state.

Abstract: An unmanned aerial vehicle is described and includes a computer carried by the unmanned aerial vehicle to control flight of the unmanned aerial vehicle and at least one sensor. The unmanned aerial vehicle is caused to fly to a specific location within a facility, where the unmanned aerial vehicle enters a hover mode, where the unmanned aerial vehicle remains in a substantially fixed location hovering over the specific location within the facility and sends raw or processing results of sensor data from the sensor to a remote server system.