Amazon Technologies, Inc.

Varun Chadha

Yongjun Wu

Sitaraman Ganapathy

Vasanthakumar Soundararajan

Abhishek Kumar

Kyle Bradley Koceski

Shyam Krishnan Moni

Systems and methods described herein provide for novel boundary generation features for interleaving additional content into media content. Media content may be received which includes a video and audio portion. An unencrypted encode of the video portion may be generated. A first set of time stamps for the video portion may be generated using a computer vision algorithm. A second set of time stamps for the video portion may be generated for identifying IDR frames using a first algorithm. A third set of time stamps may be generated to serve as boundaries for interleaving additional content into the media content based on a priority algorithm that uses the first set of time stamps and the second set of time stamps. The video portion may be encoded using the third set of time stamps to determine the IDR frames for the media content.

Generating boundary points for media content

Some implementations include methods for replacing advertisements that have been inserted into a manifest associated with live streaming and may include receiving a request for a manifest from a player to enable a user to watch a live stream of an event. The request may be received at a first time when the live stream is already in progress. The manifest may be generated to include content fragments associated with the live event and advertisements based on first set of selection criteria. The manifest may be sent to the player to enable the user to watch the live stream. A seek back request may be received from the player to enable the user to watch the live stream at a second time earlier than the first time. The manifest may be updated to replace at least one advertisement with a replacement advertisement selected based on second set of selection criteria.

Just-in-time dynamic ad insertion (DAI)

Growing community of inventors

Seattle, WA, United States of America

Varun Chadha