Amazon Technologies, Inc.

Nitin Singhal

Andrea Olgiati

Vasant Manohar

Jonathan Hedley

Gowtham Jeyabalan

Stefano Stefani

Gaurav D Ghare

Roman Goldenberg

Joseph Patrick Tighe

Ranju Das

Vivek Bhadauria

Shubham Chandra Gupta

Davide Modolo

Stephen Gould

Praveen Kumar Udayakumar

Jiazhi Ou

Yuri Natanzon

Renjun Zheng

Mark Hawley Yang

Kuang Han

Techniques for analyzing stored video upon a request are described. For example, a method of receiving a first application programming interface (API) request to analyze a stored video, the API request to include a location of the stored video and at least one analysis action to perform on the stored video; scheduling a job for the first API request using a global scheduler, the global scheduler to schedule, based at least in part on available bandwidth of processing components including a segmenter, a chunk processor, and a reducer, at least one job queue associated at least one of the processing components; accessing the location of the stored video to retrieve the stored video; segmenting the accessed video into chunks; processing each chunk with a chunk processor to perform the at least one analysis action, each chunk processor to utilize at least one machine learning model in performing the at least one analysis action; joining the results of the processing of each chunk to generate a final result; storing the final result; and providing the final result to a requestor in response to a second API request is described.

Per component schedulers making global scheduling decision

Technologies are disclosed for efficiently identifying objects in videos using deep neural networks and motion information. Using the disclosed technologies, the amount of time required to identify objects in videos can be greatly reduced. Motion information for a video, such as motion vectors, are extracted during the encoding or decoding of the video. The motion information is used to determine whether there is sufficient motion between frames of the video to warrant performing object detection on the frames. If there is insufficient movement from one frame to a subsequent frame, the subsequent frame will not be processed to identify objects contained therein. In this way, object detection will not be performed on video frames that have changed minimally as compared to a previous frame, thereby reducing the amount of time and the number of processing operations required to identify the objects in the video.

Efficient identification of objects in videos using motion information

Techniques for analyzing stored video upon a request are described. For example, a method of receiving a first application programming interface (API) request to analyze a stored video, the API request to include a location of the stored video and at least one analysis action to perform on the stored video; accessing the location of the stored video to retrieve the stored video; segmenting the accessed video into chunks; processing each chunk with a chunk processor to perform the at least one analysis action, each chunk processor to utilize at least one machine learning model in performing the at least one analysis action; joining the results of the processing of each chunk to generate a final result; storing the final result; and providing the final result to a requestor in response to a second API request is described.

Analysis of video content

A system processes a video stream to identify an activity performed by a subject. The system examines the video segment for the presence or absence of an object that is correlated with the activity. If the object is found, and if the object is positively correlated with the detected activity, a confidence level associated with detecting the activity is increase. If the object is found, and if it is negatively correlated with the activity, the confidence level is decreased. In other examples, absence of an object may be positively or negatively correlated with the activity. In one embodiment, the system may detect a number of alternative activities, and may use the presence or absence of various objects to determine the most likely activity being performed. In another embodiment, the system may use the presence of various objects to narrow the set of activities that are searched for.

Video analysis method and system

Growing community of inventors

Mercer Island, WA, United States of America

Nitin Singhal