Computer Vision System for Sub-Room Space Use Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computer vision systems lack the capability to provide identification and quantification of space use at a sub-room level of granularity while maintaining privacy and security, and they often rely on video streaming which raises privacy concerns.
Innovation Solution
A computer vision system utilizing an embedded board powered by an Nvidia GPU and quad ARM CPU processor that captures image frames, performs real-time parallel computations, and uses an open-source neural network like YOLOv2 for object detection and proximity-based tracking, without streaming video, storing only processed data to maintain privacy and security.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If video streaming is used to capture and transmit camera data, then real-time detection and analysis can be achieved, but privacy and security concerns arise due to continuous video transmission
Solution Approach 1:
The system extracts only the essential detection data (object presence, location, and basic attributes) from the video stream and transmits only this processed information to the server, rather than transmitting the entire video stream. This extraction approach maintains real-time detection capability while eliminating privacy risks associated with continuous video transmission.
Solution Approach 2:
The embedded device acts as an intermediary between the camera and the server, performing local processing of video data before transmission. This intermediary function allows the system to maintain real-time detection while filtering out sensitive information, thus resolving the contradiction between speed and privacy.
2Loss of information
If complete video streams are transmitted to the server for analysis, then comprehensive analysis can be performed, but network bandwidth and server processing resources are consumed
Solution Approach 1:
The detection task is segmented between the embedded device and the server. The embedded device performs initial detection and extracts key features, transmitting only this processed information to the server. This segmentation reduces network bandwidth consumption and server processing requirements while maintaining detection information completeness.
Solution Approach 2:
The embedded device performs preliminary detection and data extraction before transmitting information to the server. This preliminary action reduces the amount of data that needs to be transmitted and processed by the server, thereby reducing network and server resource consumption while preserving essential detection information.
3Measurement precision
If high-resolution video is captured and processed, then detection accuracy is improved, but computational requirements and processing time increase
Solution Approach 1:
The system applies local quality processing by performing detection on regions of interest within the video frame rather than processing the entire high-resolution image uniformly. This approach maintains detection accuracy for relevant objects while reducing overall computational complexity.
Solution Approach 2:
The embedded device performs partial processing of video data, focusing computational resources on detecting and extracting key features rather than processing all video data in full detail. This partial action approach maintains sufficient detection accuracy while significantly reducing computational requirements.
Data Source
AI summary
A computer vision system includes a camera that captures a plurality of image frames in a target field. A user interface is coupled to the camera. The user interface is configured to perform accelerated parallel computations in real-time on the plurality of image frames acquired by the camera. The system provides identification and quantification of space use at a sub room level of granularity.