# US 2023/0412825 A1 · Video size reduction by reconstruction

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2023/0412825 A1
- Document ID: US20230412825A1
- Status: Published application
- Kind: A1
- Application number: 17/844,152
- Filed: 2022-06-20
- Published: 2023-12-21
- Earliest claimed: 2022-06-20
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Sathya Santhar; Sarbajit K. Rakshit; Sridevi Kannan; Samuel Mathew Jawaharlal
- Web page: https://patentgrep.com/patent/US20230412825A1

## CPC

- H04N 19/42: characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation (H04N19/635 takes precedence)
- H04N 21/23418: {involving operations for analysing video streams, e.g. detecting features or characteristics (television picture signal circuitry for scene change detection H04N5/147; filtering for image enhancement G06T5/00; methods or arrangements for recognising scenes G06V20/00; arrangements characterised by components specially adapted for monitoring, identification or recognition of video in broadcast systems H04H60/59)}
- G06N 3/045: Combinations of networks
- G06N 3/0475: Generative networks
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06N 3/094: Adversarial learning
- H04N 19/132: Sampling, masking or truncation of coding units, e.g. adaptive resampling, frame skipping, frame interpolation or high-frequency transform coefficient masking
- H04N 19/154: Measured or subjectively estimated visual quality after decoding, e.g. measurement of distortion (use of rate-distortion criteria H04N19/147)
- H04N 19/167: Position within a video image, e.g. region of interest [ROI]
- H04N 19/172: the region being a picture, frame or field
- H04N 19/503: involving temporal prediction (adaptive coding with adaptive selection between spatial and temporal predictive coding H04N19/107; adaptive coding with adaptive selection among a plurality of temporal predictive coding modes H04N19/109)
- H04N 19/587: involving temporal sub-sampling or interpolation, e.g. decimation or subsequent interpolation of pictures in a video sequence
- H04N 19/593: involving spatial prediction techniques
- H04N 19/85: using pre-processing or post-processing specially adapted for video compression
- H04N 19/90: using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
- H04N 21/8456: {by decomposing the content in the time domain, e.g. in time segments}
- H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A video file is detected by a computer system. The video file that is to be provided to one or more client devices. The video file contains a video stream that includes a plurality of video images. A first video image of the plurality of video images is reconstructed based on a first machine learning technique. The first machine learning technique is based on one or more video images that occur temporally before the first video image in the video stream. A reconstruction status of the first video image of the plurality of video images is identified based on the video file and based on a second machine learning technique. An altered video file is generated in response to the reconstruction status and based on the video file.

## Claims

### Claim 1

A method comprising:
detecting, by a computer system, a video file that is to be provided to one or more client devices, the video file containing a video stream that includes a plurality of video images;
reconstructing, based on a first machine learning technique, a first video image of the plurality of video images, the first machine learning technique based on one or more video images that occur temporally before the first video image in the video stream;
identifying, based on the video file and based on a second machine learning technique, a reconstruction status of the first video image of the plurality of video images; and
generating, in response to the reconstruction status and based on the video file, an altered video file.

### Claim 2 (depends on 1)

The method of claim 1, wherein the method further comprises:
providing the altered video file to a first client device;
providing the first machine learning technique to the first client device;
causing the first client device to reconstruct, based on the first machine learning technique, the video file from the altered video file.

### Claim 3 (depends on 2)

The method of claim 2, wherein the second machine learning technique is a classifier neural network.

### Claim 4 (depends on 1)

The method of claim 1, wherein the first machine learning technique is a generative adversarial network.

### Claim 5 (depends on 1)

The method of claim 1, wherein:
the reconstruction status is that a reconstruction of the first video image of the plurality of video images is less than a quality threshold, and
the first video image is located in a first position in the video file.

### Claim 6 (depends on 5)

The method of claim 5, wherein the generating the altered video file comprises:
generating reconstruction metadata that indicates the reconstructions status of the first video image is less than the quality threshold; and
tagging the first video image with the reconstruction metadata.

### Claim 7 (depends on 5)

The method of claim 5, wherein the generating the altered video file comprises:
generating, based on a lossless compression technique, a placeholder video image for the first video image;
removing the first video image from the video file; and
inserting the placeholder video image into the first position in the video file.

### Claim 8 (depends on 5)

The method of claim 5, wherein the method further comprises:
generating, based on a compression technique and from the first video image, a compressed first video image;
reconstructing, based on a first machine learning technique and based on the compressed first video image, the first video image of the plurality of video images;
identifying, based on the video file and based on a second machine learning technique, a second reconstruction status of the first video image of the plurality of video images.

### Claim 9 (depends on 1)

The method of claim 1, wherein:
the reconstruction status is that a reconstruction of a first video image of the plurality of video images is greater than a first quality threshold,
the first video image is located in a first position in the video file, and
the generating the altered video file comprises:
generating a placeholder video image for the first video image;
removing the first video image from the video file; and
inserting the placeholder video image into the first position in the video file.

### Claim 10 (depends on 9)

The method of claim 9, wherein the placeholder video image is generated based on a compression technique.

### Claim 11 (depends on 9)

The method of claim 9, wherein the placeholder video image includes reconstruction metadata that indicates the reconstructions status of the first video image is greater than the first quality threshold.

### Claim 12 (depends on 11)

The method of claim 11, wherein the placeholder video image does not contain any image data.

### Claim 13 (depends on 1)

The method of claim 1, wherein the first video image is a video frame compressed with an intraframe compression technique.

### Claim 14 (depends on 1)

The method of claim 1, wherein the first video image is a video frame compressed with an interframe compression technique.

### Claim 15 (depends on 1)

The method of claim 1, wherein the method further comprises:
compressing, based on a file compression technique, the altered video file.

### Claim 16

A system, the system comprising:
a memory, the memory containing one or more instructions; and
a processor, the processor communicatively coupled to the memory, the processor, in response to reading the one or more instructions, configured to:
detect, by a computer system, a video file that is to be provided to a plurality of client devices, the video file containing a video stream that includes a plurality of video images;
reconstruct, based on a first machine learning technique, a first video image of the plurality of video images, the first machine learning technique based on one or more video images that occur temporally before the first video image in the video stream;
identify, based on the video file and based on a second machine learning technique, a reconstruction status of the first video image of the plurality of video images; and
generate, in response to the reconstruction status and based on the video file, an altered video file.

### Claim 17

A computer program product, the computer program product comprising:
one or more computer readable storage media; and
program instructions collectively stored on the one or more computer readable storage media, the program instructions configured to:
detect, by a client device, an altered video file to be played by the client device;
retrieve, in response to the altered video file, a first machine learning technique; and
generate, based on the first machine learning technique, a video file from the altered video file, the video file containing a video stream that includes a plurality of video images.

### Claim 18 (depends on 17)

The computer program product of claim 17, wherein the detecting the altered video comprises:
scanning the altered video file for metadata that describes the altered video file; and
identifying, based on the scanning, reconstruction metadata that indicates a reconstruction status of a first video image in the plurality of video images.

### Claim 19 (depends on 18)

The computer program product of claim 18, wherein the generating the video file comprises:
reconstructing, based on the first machine learning technique, the first video image from one or more previous video images that occur just before the first video image in the video stream.

### Claim 20 (depends on 19)

The computer program product of claim 19, wherein
the first video image is a lossy compressed image, and
the reconstructing is also based on the lossy compressed image.
