Amazon Technologies, Inc.

Eduard Oks

Matthew Alan Townsend

Apoorv Chaudhri

Ambrish Tyagi

Noam Sorek

Rohith Mysore Vijaya Kumar

Yadunandana Nagaraja Rao

Ianir Ideses

Moshe Bouhnik

Dotan Kaufman

Mark Kliger

Sharon Alpert

Ido Yerushalmy

Alexander Lorbert

Eli Alshan

Matan Goldman

Imry Kissos

Omer Meir

Lior Fritz

Konstantin Kraimer

Liza Potikha

Eran Borenstein

Arik Poznanski

Guy Adam

Assaf Neuberger

Hilit Unger

Bar Hilleli

Yaar Harari

Video output is generated based on first video data that depicts the user performing an activity. Poses of the user during performance of the activity are compared with second video data that depicts an instructor performing the activity. Corresponding poses of the user's body and the instructor's body may be determined through comparison of the first and second video data. The video data is used to determine the rate of motion of the user and to generate video output in which a visual representation of the instructor moves at a rate similar to the that of the user. For example, video output generated based on an instructional fitness video may be synchronized so that movement of the presented instructor matches the rate of movement of the user performing an exercise, improving user comprehension and performance.

System for synchronizing video output based on detected behavior

Devices, systems and methods are disclosed for improving story assembly and video summarization. For example, video clips may be received and a theme may be determined from the received video clips based on annotation data or other characteristics of the received video data. Individual moments may be extracted from the video clips, based on the selected theme and the annotation data. The moments may be ranked based on a priority metric corresponding to content determined to be desirable for purposes of video summarization. Select moments may be chosen based on the priority metric and a structure may be determined based on the selected theme. Finally, a video summarization may be generated using the selected theme and the structure, the video summarization including the select moments.

Video ingestion and clip creation

Devices and techniques are generally described for three dimensional mesh generation. In various examples, first two-dimensional (2D) image data representing a human body may be received from a first image sensor. Second 2D image data representing the human body may be received from a second image sensor. A first pose parameter and a first shape parameter may be determined using a first three-dimensional (3D) mesh prediction model and the first 2D image data. A second pose parameter and a second shape parameter may be determined using a second 3D mesh prediction model and the second 2D image data. In various examples, an updated 3D mesh prediction model may be generated from the first 3D mesh prediction model based at least in part on a first difference between the first pose parameter and the second pose parameter and a second difference between the first shape parameter and the second shape parameter.

Multi-view three-dimensional mesh generation

Systems for assisting a user in performance of a meditation activity or another type of activity are described. The systems receive user input and sensor data indicating physiological values associated with the user. These values are used to determine a recommended type of activity and a length of time for the activity. While the user performs the activity, sensors are used to measure physiological values, and an output that is provided to the user is selected based on the measured physiological values. The output may be selected to assist the user in reaching target physiological values, such as a slower respiration rate. After completion of the activity, additional physiological values are used to determine the effectiveness of the activity and the output that was provided. The effectiveness of the activity and the output may be used to determine future recommendations and future output.

Interface for guided meditation based on user interactions

A first user generates video data for performance of an activity, such as a fitness exercise, by performing the activity in front of a camera. Based on the video data, the amount of movement of different parts of the first user's body is determined. Data representing the position of the first user over time is generated. The data may take the form of a function or a signal that is based on the function. The locations of body parts that move significantly are prioritized over other body parts when determining this data. At a subsequent time, a second user performs the activity. The number of times the second user completes the activity is counted by determining the number of times the second user reaches a position corresponding to a maximum value in the data representing the position of the first user.

System for learning and tracking performance of activities

First image data representing a first human wearing a first article of clothing may be received. The first image data, when rendered on a display, may include a first photometric artifact. A first generator network may be used to generate second image data from the first image data. The first photometric artifact may be removed from the second image data. A second generator network may be used to generate third image data from the second image data, the third image data representing the first human in a different pose relative to the first image data. Fourth image data representing the first article of clothing segmented from the first human may be generated and displayed on a display.

Catalog normalization and segmentation for fashion images

Techniques are generally described for performing outfit recommendation using a recurrent neural network. In various examples, a computing device may receive a first state vector representing an outfit comprising at least one article of clothing. First image data depicting a second article of clothing of a first clothing category may be received. A recurrent neural network may generate a first output feature vector based on the first state vector, the first image data and the first clothing category. The first output feature vector may be compared to other feature vectors representing other articles of clothing in the first category to determine distances between the first output feature vector and the other feature vectors. A set of articles of clothing may be recommended based on the distances between the first output feature vector and the other feature vectors.

Outfit recommendation using recurrent neural networks

A system and method for determining video clips including interesting content from video data. The system may receive annotation data identifying time and positions corresponding to objects represented in the video data and the system may determine priority metrics associated with each of the objects. By associating the priority metrics with the time and positions corresponding to the objects, the system may generate a priority metric map indicating a time and position of interesting moments in the video data. The system may generate moments and/or video clips based on the priority metric map. The system may determine a time (e.g., video frames) and/or space (e.g., pixel coordinates) associated with the moments/video clips and may simulate camera motion such as panning and/or zooming with the moments/video clips. The system may generate a Master Clip Table including the moments, video clips and/or annotation data associated with the moments/video clips.

Cataloging video and creating video summaries

Devices, systems and methods are disclosed for reducing a perceived latency associated with uploading and annotating video data. For example, video data may be divided into video sections that are uploaded individually so that the video sections may be annotated as they are received. This reduces a latency associated with the annotation process, as a portion of the video data is annotated before an entirety of the video data is uploaded. In addition, the annotation data may be used to generate a master clip table and extract individual video clips from the video data.

A computer-implemented method includes receiving first pose data for a first human represented in a first image, receiving second pose data for a second human represented in a second image, receiving first semantic segmentation data for the first image, and receiving second semantic segmentation data for the second image. A pose-aligned second image can be generated by modifying the second image based on the first pose data, the second pose data, the first semantic segmentation data, and the second semantic segmentation data. A mixed image can be determined by combining pixel values from the first image and pixel values of the pose-aligned second image according to mask data. In some embodiments, the mixed image includes a representation of an outfit that includes first clothing represented in the first image and second clothing represented in the second image.

Growing community of inventors

Holon, Israel

Eduard Oks