18563734. VIDEO COMPRESSION USING OPTICAL FLOW simplified abstract (Google LLC)

VIDEO COMPRESSION USING OPTICAL FLOW

Organization Name

Google LLC

Inventor(s)

George Dan Toderici of Mountain View CA (US)

Eirikur Thor Agustsson of Zurich (CH)

Fabian Julius Mentzer of Zurich (CH)

David Charles Minnen of Mountain View CA (US)

Johannes Balle of San Francisco CA (US)

Nicholas Johnston of San Jose CA (US)

VIDEO COMPRESSION USING OPTICAL FLOW - A simplified explanation of the abstract

This abstract first appeared for US patent application 18563734 titled 'VIDEO COMPRESSION USING OPTICAL FLOW

Simplified Explanation: The patent application describes a method for compressing video data using optical flow prediction and autoencoder neural networks.

• Optical flow prediction network generates optical flow between sequential frames.
• First autoencoder neural network generates predicted optical flow between frames.
• Warping and blurring operations are applied to obtain a predicted reconstruction of frames.

Key Features and Innovation:

• Use of optical flow prediction network for video compression.
• Integration of autoencoder neural networks for frame reconstruction.
• Application of warping and blurring operations for improved compression.

Potential Applications:

• Video streaming services
• Video surveillance systems
• Video editing software

Problems Solved:

• Efficient video data compression
• Reduced bandwidth requirements for video transmission
• Improved video quality at lower data rates

Benefits:

• Reduced storage space for video files
• Faster video transmission over networks
• Enhanced video compression without significant loss of quality

Commercial Applications: The technology can be utilized in various industries such as entertainment, security, and communications for efficient video data compression and transmission.

Prior Art: Researchers can explore prior art related to optical flow prediction, autoencoder neural networks, and video compression techniques.

Frequently Updated Research: Stay updated on advancements in optical flow prediction, neural network algorithms, and video compression technologies for potential improvements in the field.

Questions about Video Compression: 1. How does optical flow prediction contribute to video compression? 2. What are the advantages of using autoencoder neural networks in video data compression?

Original Abstract Submitted

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for compressing video data. In one aspect, a method comprises: receiving a video sequence of frames; generating, using a flow prediction network, an optical flow between two sequential frames, wherein the two sequential frames comprise a first frame and a second frame that is subsequent the first frame; generating from the optical flow, using a first autoencoder neural network: a predicted optical flow between the first frame and the second frame; and warping a reconstruction of the first frame according to the predicted optical flow and subsequently applying a blurring operation to obtain an initial predicted reconstruction of the second frame.