A record-and-replay control device includes a video data acquisition unit configured to acquire video data from an imaging unit to capture an image of outside of a vehicle; an event detector configured to detect an event of the vehicle; an event detection direction acquisition unit configured to acquire an event detection direction of the event; a recording controller configured to store video data caused by the detected event as event recording data with a first angle of view; a replay controller configured to replay selected event recording data; and a display controller configured to control a display unit to display, out of thumbnail images of the stored event recording data, a thumbnail image indicating the event recording data to which the event detection direction is correlated, as a thumbnail image obtained by adopting a second angle of view with enlargement in the event detection direction.

Techniques and systems are provided for generating media content. For example, a server computer can detect a trigger from a device located at a site. The trigger is associated with an event at the site. The server computer can obtain media segments of media captured by a plurality of media capture devices located at the site. At least one of the media segments corresponds to the detected trigger. The server computer can determine one or more quality metrics of a media segment based on a first motion of an object captured in the media segment and/or a second motion of a media capture device used to capture the media segment. A subset of media segments can be selected from the obtained media segments based on quality metrics determined for the obtained media segments. A collection of media segments including the subset of media segments can then be generated.

A method for video recognition and related products are provided. The method includes the following. An original set of clip descriptors is obtained by providing multiple clips of a video as an input of a 3D CNN of a neural network, where the neural network includes the 3D CNN and at least one first fully connected layer, and each of the multiple clips includes at least one frame. An attention vector corresponding to the original set of clip descriptors is determined. An enhanced set of clip descriptors is obtained based on the original set of clip descriptors and the attention vector. The enhanced set of clip descriptors is input into the at least one first fully connected layer and video recognition is performed based on an output of the at least one first fully connected layer.

A system for determining whether shoppers are eligible for frictionless checkout is disclosed. The system has a processor that obtains image data captured using image sensors positioned in a retail store. The processor analyzes the image data to identify at least one shopper at one or more locations of the retail store. The processor detects, based on the analysis of the image data, at least one product interaction event associated with an action of the at least one shopper at the one or more locations of the retail store. Further, based on the detected at least one product interaction event, the processor determines whether the at least one shopper is eligible for frictionless checkout. In response to a determination that the at least one shopper is ineligible for frictionless checkout, the processor causes delivery of an indicator that the at least one shopper is ineligible for frictionless checkout.

This disclosure describes interactive data processing systems configured to facilitate selection by a human associate of tentative results generated by an automated system from sensor data. In one implementation, an event may take place in a materials handling facility. The event may comprise a pick or place of an item from an inventory location, movement of a user, and so forth. The sensor data associated with the event is processed by an automated system to determine tentative results associated with the event. In some situations, an uncertainty may exist as to which of the tentative results accurately reflects the actual event. The system may then determine whether the event is to be merged with one or more temporally and spatially proximate events and, if so, the sensor data and tentative results for the merged event is sent to a human associate. The associate may select one of the tentative results.

Generally, a system and method for an automatic detection of referee's decisions during a ball-game match are provided. The method may include receiving a plurality of images of a ball-game field generated during the ball-game match; determining, based on predetermined ball-game rules, a first subset of images of the plurality of images representing a first event that is suspected as a specified rule-based event; determining, based on the predetermine ball-game rules, a second subset of images of the plurality of images that represents a second event, wherein the second event is subsequent to the specified rule-based event according to the predetermined ball-game rules; and analyzing, based on the predetermined ball-game rules, the images of the second subset and further determining, based on the analysis thereof, a referee's decision concerning whether the first even is the specified rule-based event.

A computing system retrieves ball-by-ball data for a plurality of sporting events. The computing system generates a trained neural network based on ball-by-ball data supplemented with ball-by-ball data with ball-by-ball match context features and personalized embeddings based on a batsman and a bowler for each delivery. The computing system receives a target batsman and a target bowler for a pitch to be delivered in a target event. The computing system identifies target ball-by-ball data for a window of pitches preceding the to be delivered pitch. The computing system retrieves historical ball-by-ball data for each of the target batsman and the target bowler. The computing system generates personalized embeddings for both the target batsman and the target bowler based on the historical ball-by-ball data. The computing system predicts a shot type for the pitch to be delivered based on the target ball-by-ball data and the personalized embeddings.

A display assistant device comprises a display, a camera, a speaker, microphones, a processor and memory. The memory stores programs comprising instructions that, when executed by the processor, enable a plurality of modes of the display assistant device. The modes include a monitoring mode and an assistant mode. In the monitoring mode, the device is configured to perform a remote monitoring function in which first video captured by the camera is streamed to a remote server system for monitoring uses. The monitoring uses include transmission of the first video to remote client devices authorized to access the first video. In the assistant mode, the device is configured to perform a second plurality of functions that excludes the monitoring function and includes a video communication function in which second video captured by the camera is transmitted to second devices participating in a video communication with a first user of the device.

Introduced here are health management platforms able to monitor changes in the health state of a subject based on the context of digital activities performed by, or involving, the subject. Initially, a health management platform can identify a physiological response by examining physiological data associated with a subject. Then, the health management platform can identify a stimulus presented by an electronic device that provoked the physiological response by examining contextual data associated with the subject. The contextual data may be in the form of a screenshot of a computer program in use by the subject during the physiological response. In some embodiments, the health management platform prompts the subject to specify whether the physiological response is a positive physiological response that resulted in an upward shift in health or a negative physiological response that resulted in a downward shift in health.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for managing virtual surveillance windows for video surveillance. The methods, systems, and apparatus include actions of obtaining an original video, generating a downscaled video from the original video, detecting a first event at a location from the downscaled video using a first classifier, generating a windowed video from the original video based on the location, detecting a second event from the windowed video, and performing an action in response to detecting the second event.