Temporal Resampling Modes for Low-Bitrate Machine Vision Video

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing video coding technologies struggle to efficiently manage bitrate requirements and adapt to the differing video quality needs of machine vision tasks, leading to inefficiencies in encoding and decoding processes.

Innovation Solution

Implementing temporal resampling and restoration modes, along with post-filtering techniques, to optimize video encoding and decoding for machine vision applications, allowing for adaptive sampling and improved video quality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If video is encoded using traditional compression techniques, then bitrate requirements are reduced, but video quality for machine vision tasks deteriorates

Engineering Contradiction:

ImprovebitrateVSAvoidvideo quality

Core Design Contradiction:

Quantity of substanceVSMeasurement precision

Solution Approach 1:

The patent changes the temporal sampling parameters by introducing temporal resampling modes (interpolation, extrapolation, and mixed modes) that operate at different ratios (2x, 3x, 4x, etc.). This allows the system to reconstruct video frames at various temporal resolutions, optimizing the balance between bitrate reduction and maintaining sufficient quality for machine vision tasks.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent implements dynamic temporal resampling where the resampling ratio and mode can vary across different video sequences, pictures, and even regions within pictures. The system adapts the temporal sampling strategy based on content characteristics, motion complexity, and specific machine vision task requirements, rather than applying a fixed compression approach.

Inventive Principle:

Principle #15Dynamics

2Productivity

If temporal resampling is applied to reduce data volume, then transmission efficiency improves, but encoding complexity increases

Engineering Contradiction:

Improvetransmission efficiencyVSAvoidencoding complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the temporal resampling process into distinct operational modes (interpolation mode, extrapolation mode, and mixed mode) that can be independently selected and applied. Each mode handles different temporal relationships between frames, allowing the system to break down the complex resampling task into manageable, standardized operations.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces temporal resampling as an intermediary process between video capture and encoding. This intermediary layer preprocesses the video stream by inserting or removing frames according to specified ratios and modes, thereby reducing the data volume that needs to be encoded and transmitted while maintaining temporal information integrity.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Measurement precision

If adaptive temporal sampling is implemented, then video quality for machine vision tasks is optimized, but processing time increases

Engineering Contradiction:

Improvevideo qualityVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies partial temporal resampling by allowing selective application of different resampling ratios to different regions or time periods within the video stream. Rather than uniformly applying high-resolution temporal sampling throughout, the system uses lower resampling ratios where sufficient and higher ratios only where needed for machine vision tasks, reducing overall processing time.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS20260019604A1Methods and devices for temporal resampling modes and temporal resampling post filtering

Publication Date: 2026.01.15 TENCENT AMERICA LLC

AI summary

This disclosure relates generally to video coding/decoding and particularly for signaling in temporal resampling modes and/or post filtering in video coding and/or decoding systems. One method includes obtaining a coded video bitstream; determining, from the coded video bitstream, a sequence-level temporal restoration flag for a picture sequence; when the sequence-level temporal restoration flag indicates that temporal restoration is enabled for the picture sequence, determining, from the coded video bitstream, a temporal restoration mode for the picture sequence; when the temporal restoration mode indicates an interpolation mode, determining, from the coded video bitstream, an interpolation ratio index indicating a temporal resampling ratio; when the temporal restoration mode indicates an extrapolation mode, determining, from the coded video bitstream, an extrapolation ratio index indicating a temporal resampling ratio; and decoding the coded video bitstream by generating temporal resampling data based on the temporal resampling ratio.