Video Frame Selection Using Object Classification for Storage Reduction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The large size of surveillance video data requires significant storage space, with many frames containing no moving objects, leading to inefficient use of storage resources.

Innovation Solution

A method and system for video processing that determines a target bounding box in an image frame, classifies the subject within the box using a recognition model, and selectively saves frames based on classification results and conditions, such as historical processing counts and pixel differences, to reduce unnecessary storage.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If all video frames are stored, then complete surveillance coverage is maintained, but storage space requirements increase significantly

Engineering Contradiction:
Improvesurveillance coverageVSAvoidstorage space
Core Design Contradiction:
ReliabilityVSVolume of stationary object

Solution Approach 1:

The patent extracts and identifies moving objects from video frames using object detection algorithms. Only frames containing detected moving objects are selected for storage, while frames without moving objects are discarded. This extraction principle directly resolves the contradiction by separating useful surveillance information from redundant static frames, maintaining reliability for actual events while dramatically reducing storage requirements.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent discards video frames that do not contain moving objects based on detection results. By systematically discarding redundant frames and retaining only those with meaningful content, the system achieves efficient storage utilization while preserving all potentially important surveillance data for recovery and analysis when needed.

Inventive Principle:
Principle #34Discarding and recovering

2Measurement precision

If object detection and classification are performed on every frame, then accurate identification of important frames is achieved, but processing time and computational resources increase

Engineering Contradiction:
Improveframe selection accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary object detection and classification on video frames to identify frames containing moving objects before final storage decisions are made. This preliminary action allows the system to pre-screen frames and avoid unnecessary processing of frames that will ultimately be discarded, thereby reducing overall processing time while maintaining accurate frame selection through subsequent verification steps.

Inventive Principle:
Principle #10Preliminary action

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

Efficiently stores only relevant video frames, reducing storage requirements by eliminating frames without moving subjects, thus optimizing storage utilization.

Implementation Method 1

determining a classification result of a subject in the target bounding box by processing the target bounding box using a recognition model

Methodology Applied
Scientific EffectImage Processing:

Implementation Method 2

determining the background image by processing the video using a Gaussian background modeling algorithm

Methodology Applied
Scientific EffectGaussian background modeling:

Data Source

PatentUS12586339B2Methods and systems for video processing
Publication Date: 2026.03.24 ZHEJIANG DAHUA TECH CO LTD
  • US12586339B2 patent drawing
  • US12586339B2 patent drawing
  • US12586339B2 patent drawing

AI summary

A method and a system for video processing, including: determining a target bounding box from an image frame in a video; determining a classification result of a subject in the target bounding box by processing the target bounding box using a recognition model; and determining one or more target image frames to be saved in the video based on the classification result.