Distributed Video Preprocessing for Privacy-Safe CCTV Analytics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing CCTV systems face challenges in efficiently transmitting large image data while ensuring privacy protection, especially in public transportation, where videos are typically stored locally due to size constraints and privacy concerns.
Innovation Solution
A distributed processing method and apparatus that preprocesses video data on client devices, extracting and transmitting only necessary information such as cropped objects, grayscale conversion, and blurred videos, along with location and attribute data, using a communication unit to enhance data transmission efficiency and protect privacy by virtualizing personal identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If video data is transmitted to a centralized server for processing, then comprehensive data analysis and policy establishment are improved, but data transmission efficiency deteriorates due to large image data size and privacy protection issues
Solution Approach 1:
The system segments video data into essential information extracts (location information, attribute information, cropped videos of objects of interest) and non-essential data. Only the segmented essential information is transmitted to the server, reducing data volume while maintaining analytical value for policy establishment and incident detection.
Solution Approach 2:
The client device extracts and removes unnecessary data from the original video before transmission. Only essential information such as location coordinates, object attributes, and cropped relevant frames are extracted and transmitted to the server, significantly reducing the quantity of data transmitted while preserving analytical effectiveness.
2Measurement precision
If original video data is transmitted to ensure complete information for analysis, then analysis accuracy is improved, but privacy protection deteriorates
Solution Approach 1:
The system extracts only the necessary analytical information from video data (location, attributes, cropped frames of interest) while removing or anonymizing personally identifiable information. This extraction approach maintains analysis accuracy for detecting incidents and establishing policies while protecting individual privacy.
Solution Approach 2:
The patent introduces an information processing unit as an intermediary between the client device and server. This intermediary processes and filters data to extract only essential information needed for analysis, acting as a mediator that preserves analytical accuracy while eliminating privacy risks associated with transmitting complete video data.
3Object-affected harmful factors
If video data is stored locally at client devices, then privacy protection is improved, but data availability for centralized analysis deteriorates
Solution Approach 1:
The system segments data into sensitive information (stored locally at client devices) and analysis-ready information (transmitted to server). Local storage of original videos maintains privacy protection, while extracted essential information ensures data availability for centralized analysis and policy establishment.
Solution Approach 2:
The client device performs preliminary processing and extraction of essential information locally before transmission to the server. This preliminary action ensures that only necessary data is transmitted, maintaining privacy protection while ensuring that sufficient data is available for centralized analysis.
Data Source
AI summary
A distributed processing apparatus connected to one or more clients through a network includes a communication unit that receives, from the one or more clients, preprocessing information on a video of a target space input from the one or more clients, and location information and attribute information on one or more objects of interest included in the target space; a storage unit that stores at least one of the preprocessing information, the location information, and the attribute information; and an information processing unit that generates visualization information on the one or more objects or performs inference by a heat map or a digital twin method using the stored information.