Video Camera Object Detection Using Local Preprocessing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing camera-equipped home security systems are expensive due to the high computational cost of processing images for object detection, requiring remote servers and human intervention, and lack efficient local object detection capabilities.
Innovation Solution
A method that uses a video camera to detect and record objects within its field of view by analyzing image frames, comparing sets of frames to identify arriving or departing objects, and using a database to track and update object presence, employing techniques like block-based discrete cosine transform and color histogram comparison to filter out artifacts and determine object changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If remote servers are used to process captured images for object detection, then detection accuracy is improved, but system cost and computational overhead increase significantly
Solution Approach 1:
The patent segments the object detection task into local preprocessing at the camera device and remote processing only for detected events. The camera performs local object detection and generates event notifications, while the remote server only processes these notifications to determine final object types. This segmentation reduces the computational burden on remote servers while maintaining detection accuracy.
Solution Approach 2:
The camera device is equipped with local object detection capabilities that enable it to autonomously identify objects and generate event notifications without requiring constant remote server intervention. The system serves itself by performing initial detection locally, reducing dependency on expensive remote processing resources.
2Reliability
If comprehensive image analysis is performed to robustly determine object presence, then detection reliability is improved, but computational cost increases
Solution Approach 1:
The system performs preliminary object detection and event generation at the camera device before transmitting data to the remote server. This preliminary action filters out unnecessary computations at the server level, as only events requiring further analysis are processed remotely, thereby reducing overall computational cost while maintaining reliability.
Solution Approach 2:
Instead of performing complete image analysis on all captured frames, the system applies partial analysis only to frames containing detected events. This selective processing approach reduces computational cost by avoiding excessive analysis of irrelevant frames while maintaining detection reliability for actual events.
3Productivity
If local object detection is implemented to reduce computational costs, then system efficiency is improved, but detection capability may be reduced
Solution Approach 1:
The detection system is segmented into two levels: local detection at the camera for event generation, and remote detection for final object classification. This segmentation allows the system to maintain high detection capability by leveraging remote server resources for complex classification tasks while achieving improved efficiency through local preprocessing.
Solution Approach 2:
Event notifications serve as an intermediary between local and remote processing. The camera generates these notifications based on local detection, which then trigger remote server processing. This intermediary mechanism ensures that local detection capabilities are sufficient for event identification while remote resources enhance overall detection capability.
Data Source
AI summary
A method for processing arrival or removal of packages within the field of view of a video camera includes providing a database for recording packages placed in the field of view. Based on real-time analysis of successive image frames in the camera, a human person's entry and exit from the field of view of the camera is also detected. Delivery or removal of objects is recorded in the database. In one embodiment, the method also determines whether or not a newly arrived package is placed alongside or on top of an existing package.