Video Analytics & AI Monitoring Flashcards
7 cards from real CSP practice questions. Tap to flip, then mark Knew It or Still Learning โ missed cards come back until you master them.
Read the first 7 Video Analytics & AI Monitoring flashcards as text
Which video analytic capability is specifically designed to track the movement path of an individual across multiple non-overlapping cameras?
Answer: Multi-camera re-identification (Re-ID)
Multi-camera re-identification matches the appearance of individuals across different camera views without requiring continuous optical coverage between them.
For a CSP professional, what is the main legal concern when deploying AI-based biometric surveillance in the US?
Answer: State biometric privacy laws such as Illinois BIPA
Laws like Illinois' Biometric Information Privacy Act (BIPA) impose strict requirements on collecting, storing, and using biometric identifiers such as facial templates.
What does 'Intersection over Union' (IoU) measure in object detection evaluation?
Answer: The degree of overlap between predicted and ground-truth bounding boxes
IoU calculates the ratio of the overlapping area to the combined area of the predicted and actual bounding boxes, indicating detection accuracy.
In a smart city surveillance deployment, which AI analytic would BEST support traffic incident detection on a highway?
Answer: Stopped vehicle and wrong-way driver detection
Stopped vehicle and wrong-way driver analytics specifically identify high-risk traffic events on roadways that require immediate emergency response.
What is the primary security risk of connecting AI video analytics servers directly to the public internet without network segmentation?
Answer: Exposure to unauthorized access and potential manipulation of analytics rules
Without network segmentation, analytics servers are vulnerable to external attacks that could disable alerts, manipulate detection rules, or exfiltrate sensitive data.
Which approach BEST mitigates AI surveillance bias against individuals with darker skin tones identified in facial recognition studies?
Answer: Ensuring training datasets are demographically diverse and balanced
Bias in facial recognition largely stems from underrepresentation of certain demographics in training data; diverse, balanced datasets significantly reduce disparate error rates.
A CSP professional must justify the ROI of an AI video analytics upgrade to management. Which metric is MOST persuasive?
Answer: Reduction in security incidents and guard labor costs achieved post-deployment
Demonstrating measurable decreases in incidents and reduced staffing costs directly ties the AI investment to quantifiable business and safety outcomes.