Detection Dashboard

Connecting… 0 results
👥 0
📷

No detections yet — analyse an image in the WPF app to see results here.

No acknowledged detections yet. Use the ✔ / ✘ buttons on a detection card.

Total Detections
Acknowledged
Positive
False / Negative
Avg Response Time
Detections & Acknowledgements per Day
Top Incident Types
Top Servers (Incidents/camera)
Top Servers
Top Cameras
Average Response Time (minutes)
Alerts by Hour of Day
Select a camera above to view hourly breakdown

Capture Errors

Date / Time Server Name Camera Name Base Server Event ID Reason
Loading…

Roof Alerts

Event Name Property Name Base Server Status Last Active Actions
Loading…

Package Assist

Server Health

Server Name Cameras Last Alert Total Alerts Alerts (7d) Avg/Day False % Most Active Camera
Loading…

Operator Productivity

User Cohort False Positive Pos 30d Pos 7d Pos/wk TTA Last pos Shifts Pos/shift Δ 4-wk Shift hrs

Week-over-week: positive acknowledgments per shift (last 12 complete ET weeks; the in-progress week is excluded)

Each investigator is plotted against their own history; the dashed line is the cohort median for that week, so a slow week for everyone reads as a slow week rather than a decline. Non-investigators are excluded.

Activity by hour (ET, last 90 days, all acknowledgments)

AI Security Analysis

Generate an AI-powered security analysis report for one or more properties.

Generate Report

to

False-Positive Analysis

Window
Observe-only — nothing here is hidden, auto-classified, or suppressed. It shows where false positives concentrate and how the "Likely False" prediction is performing, so any future action can be earned by the numbers.

Prediction accuracy

Scored only on acknowledgements made after this shipped (older ones have no recorded prediction).

Where the false positives live

False-rate by camera & hour

Database

Total Detections
Total Acknowledged
Capture Errors
Detections (24h)
Detections (7d)
DB Size
DTU Now
DTU Peak (1h)
Workers Peak (1h)
Long-Running (>5s)
Longest Query Now
Largest Table

Local Image Storage

Local Images
Local Images Size
Disk Used
Disk Free
Oldest Unarchived

Blob Storage

Blob Images
Blob Storage Size
Oldest Blob
Newest Blob

Archive Service

Last Run Status
Last Run
Next Run
Last Archived
Last Deleted
Last Failed

Application

Uptime
Memory (Working Set)
Connected Users
Server Time (UTC)

Configuration

Map sender IP addresses to friendly server names. These names appear on detection cards instead of the raw IP address.

IP Mappings

IP AddressBase Server NameAddedActions
Loading…

SGS Secondary Password Servers

Servers listed here will use the secondary SGS password directly, skipping the primary password attempt. Add any server that consistently fails with the primary password.

IP AddressNoteAddedActions
Loading…

Detection Type Priorities

Map YOLO/Roboflow detection class names (e.g. "knife", "person", "car") to a priority (1–10). When sorting by Priority on the dashboard, alerts score by their highest priority detection class. Alerts with no class mapping fall back to person count as the priority score.

Detection ClassPriorityActions
Loading…

Incident Types

Manage the list of incident types available in the acknowledgement dropdown, and classify each as positive or negative. This list only populates the dropdown — operators still choose positive/negative when acknowledging, and existing records are unaffected.

Incident TypeClassificationExclude Images from ReportActions
Loading…

Image Archive

Archives local images to Azure Blob Storage and deletes false detection blobs per the configured retention policy. Normally runs nightly at 2 AM.

User Management

Manage who can access the Detection Dashboard.

Two-Factor Authentication

Require email verification on login for all users

To disable via SQL if locked out: UPDATE AppSettings SET Value='false' WHERE [Key]='TwoFactorEnabled'

Create New User

All Users

Username Email Role Cohort Permissions Created Last Login # Acknowledged Actions
Loading…

Login Audit Log

Time Username IP Address Result Reason
Loading…