Visitor Recognition Algorithm for Security Devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Security devices, such as video doorbells and floodlight cameras, generate unnecessary notifications when detecting frequent visitors, leading to alert fatigue as users receive unimportant alerts alongside important ones, failing to distinguish between expected and unexpected events.
Innovation Solution
Implementing a visitor recognition system that uses a prior visitor model to differentiate between known and unknown visitors by analyzing facial images and annotation data, reducing unnecessary notifications by identifying regular visitors and alerting only for unusual or unexpected events.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the security device generates notifications for every detected visitor, then all visitors are alerted to the system, but users experience alert fatigue due to unnecessary notifications from frequent visitors
Solution Approach 1:
The system performs preliminary learning during a collection period to build a prior visitor model before generating notifications. This preliminary action allows the system to distinguish between known and unknown visitors, enabling selective notification only for unexpected visitors and eliminating alert fatigue from repeated notifications to frequent visitors.
Solution Approach 2:
The system uses feedback from the collection period to update the prior visitor model continuously. By analyzing visitor patterns over time and comparing them against the model, the system dynamically adjusts its notification behavior to maintain high accuracy while reducing unnecessary alerts to users.
2Measurement precision
If the system uses facial recognition to identify visitors, then visitor identification accuracy improves, but the complexity of the system increases
Solution Approach 1:
The system extracts only the facial portion from captured images and uses this extracted feature for identification purposes. By focusing solely on facial recognition rather than analyzing entire images or using multiple recognition methods, the system achieves accurate visitor identification while minimizing computational complexity and processing requirements.
Data Source
AI summary
A system and visitor recognition method receives a first image of a detected visitor at a first location, compares the image of the detected visitor to a prior visitor model including images of prior visitors to the first location or a second location. In some embodiments, when an image of the detected visitor matches the images of the prior visitor captured at a time differing from the capture time of the image of the detected visitor, the detected visitor is determined as unknown. In some embodiments, when the image of the detected visitor matches a group of images of a prior visitor at the second location, and the number of times the prior visitor was previously detected at the second location is above a second threshold, the detected visitor is determined as known, even if the detected visitor was not detected at the first location at least a threshold number of times.