Identifying activity in an area even during periods of poor visibility using a video camera and an audio sensor are disclosed. The video camera is used to identify visible events of interest and the audio sensor is used to capture audio occurring temporally with the identified visible events of interest. A sound profile is determined for each of the identified visible events of interest based on sounds captured by the audio sensor during the corresponding identified visible event of interest. Then, during a time of poor visibility, a subsequent sound event is identified in a subsequent audio stream captured by the audio sensor. One or more sound characteristics of the subsequent sound event are compared with the sound profiles associated with each of the identified visible events of interest, and if there is a match, one or more matching sound profiles are filtered out from the subsequent audio stream.

A vehicle dash cam may be configured to execute one or more neural networks (and/or other artificial intelligence), such as based on input from one or more of the cameras and/or other sensors associated with the dash cam, to intelligently detect safety events in real-time. Detection of a safety event may trigger an in-cab alert to make the driver aware of the safety risk. The dash cam may include logic for determining which asset data to transmit to a backend server in response to detection of a safety event, as well as which asset data to transmit to the backend server in response to analysis of sensor data that did not trigger a safety event. The asset data transmitted to the backend server may be further analyzed to determine if further alerts should be provided to the driver and/or to a safety manager.

A technique for detecting a glitch in an image is provided. The technique includes providing an image to a plurality of individual classifiers to generate a plurality of individual classifier outputs and providing the plurality of individual classifier outputs to an ensemble classifier to generate a glitch classification.

A system and method for detecting false or exaggerated responses to physical contact. Image capture devices are positioned around a monitored area for capturing still or video images of an activity that include a moment of contact. Those images are synchronized to allow for composite analysis. Then object detection is performed to identify objects of interest within the images. Structural elements of the objects are then identified. Based on the structural elements, a determination is made as to whether a contact was made, for example based on whether structural elements of two nearby objects became adjacent or overlapping with one another, a force is calculated based on a speed of movement of those elements, and then machine learning is used to determine whether a response to contact was within reasonable expectations based on the amount of force and other factors.

Automated detection of features and/or parameters within an ocean environment using image data. In an embodiment, captured image data is received from ocean-facing camera(s) that are positioned to capture a region of an ocean environment. Feature(s) are identified within the captured image data, and parameter(s) are measured based on the identified feature(s). Then, when a request for data is received from a user system, the requested data is generated based on the parameter(s) and sent to the user system.

A method for action recognition in a video is described. The method includes inputting a plurality of consecutive clips divided from the video into a convolutional neural network (CNN), and obtaining a set of clip descriptors; processing the set of clip descriptors via a Bi-directional Attention mechanism, and obtaining a global representation of the video; and performing video-classification for the global representation of the video such that action recognition is achieved.

A computer-implemented method of processing video data comprising a first sequences of image frames containing a first instance of an object. The method includes isolating said first instance of the object within the first sequence of image frames, determining, using the isolated first instance of the object, first parameter values for a synthetic model of the object, modifying the first parameter values for the synthetic model of the object, rendering a modified first instance of the object using a trained machine learning model and the modified first parameter values for the synthetic model of the object, and replacing at least part of the first instance of the object within the first sequence of image frames with a corresponding at least part of the modified first instance of the object.

Systems, computer program products, and methods are described herein for real-time identification of unauthorized access. The present invention is configured to receive, from a first computing device, an indication of a first triggering activity; extract one or more first assessment vectors associated with the first triggering activity; receive, from a second computing device, an indication of a second triggering activity; extract one or more second assessment vectors associated with the second triggering activity; determine, in real-time, that the one or more first assessment vectors and the one or more second assessment vectors indicate an incidence of a misappropriate activity; dynamically generate, using an augmented reality application, an augmented reality overlay comprising at least the one or more first assessment vectors, the one or more second assessment vectors, and an indication of the incidence of the misappropriate activity.

Techniques for collecting, synchronizing, and displaying various types of data relating to a road segment enable, via one or more local or remote processors, servers, transceivers, and/or sensors, (i) enhanced and contextualized analysis of vehicle events by way of synchronizing different data types, relating to a monitored road segment, collected via various different types of data sources; (ii) enhanced and contextualized analysis of filed insurance claims pertaining to a vehicle incident at a road segment; (iii) advantageous machine learning techniques for predicting a level of risk assumed for a given vehicle event or a given road segment; (iv) techniques for accounting for region-specific driver profiles when controlling autonomous vehicles; and/or (v) improved techniques for providing a GUI to display collected data in a meaningful and contextualized manner.

A system detects a transaction outcome by obtaining video data associated with a transaction area and analyzing the video data to obtain at least one video transaction parameter concerning transactions associated with the transaction area. The transaction area can be a video count of items indicated in the video data as detected by an automated item detection algorithm applied to the video data. The system obtains at least one expected transaction parameter concerning an expected transaction that occurs in the transaction area, such as a scan count of items scanned at a point of sale terminal. The system automatically compares the video transaction parameter(s) to the expected transaction parameter(s) to identify a transaction outcome that may indicate fraudulent activity such as sweethearting in a retail environment.