Perimeter Cameras
Start on fixed views where a normal scene and expected coverage are well understood.
Camera Obstruction Detection for Video Quality Review
A blocked or altered view can leave an operator without the camera context they expect. StreamVLM™ can flag the visible quality change for a technician or control-room review.
The Visible Condition
Give the Operator a Frame to Review
The condition in scope is a camera view that is covered, blocked, strongly blurred, shifted, or otherwise visibly unusable. It is a visible scene detail, not a diagnosis, a conclusion about intent, or an automatic decision.
In StreamVLM™, an authorized operator writes a prompt such as “Alert when this camera view is obstructed.” The model evaluates selected live frames against that prompt and sends a match, camera, location, and time into the IREX event workflow for human review.
StreamVLM™ runs on a GPU node in the customer instance. Each prompt has its own confidence threshold, alert cooldown, event type, and camera selection.
This prompt-defined detector can be evaluated alongside the Camera Tamper Detection trained-module workflow. The two approaches have different frame-rate and operational-fit considerations, so a pilot should establish which is appropriate for the site.
Explore StreamVLM Prompt DetectorsWhere It Helps
Set the Prompt Where the Condition Matters
Public-Space Cameras
Let operators distinguish temporary weather effects from a persistent camera-integrity issue.
Critical-Site Entrances
Route confirmed faults to the technician or service desk responsible for that camera.
Illustrative Camera Frames
What the Operator Reviews
These illustrative frames show the kind of visible condition a prompt can surface. They are not real incidents, deployments, or proof of detection accuracy.
Camera obstruction
A Changed View
An operator compares the live scene with the expected camera coverage.
Camera obstruction
Partial Obstruction
The frame can be routed for a check without attributing the cause.
Camera obstruction
A View That Needs Attention
Site staff decide whether cleaning, repositioning, or another response is required.
From Prompt to Action
A Review Workflow for This Condition
- 01
Establish the Expected Camera View
Write a concise prompt for a camera view that is covered, blocked, strongly blurred, shifted, or otherwise visibly unusable, then select the cameras and times where the condition warrants attention.
- 02
Review the Quality Change and Nearby Conditions
Open the event with its camera, location, time, and available video context. Confirm what is visible before treating the match as actionable.
- 03
Create the Appropriate Maintenance or Integrity Task
The authorized team applies the local procedure. StreamVLM supplies a review cue; it does not decide the response.
Limits to Test
Validate It on the Actual Cameras
- Scene Context
- Weather, darkness, glare, and temporary scene changes can reduce image quality without obstruction.
- Camera Fit
- Each view needs a baseline that makes an unexpected change reviewable.
- Human Decision
- The event does not establish the cause of the degradation or identify a person.
FAQ
Can it identify who obstructed a camera?
No. The prompt concerns the visible condition of the camera view. A separate investigation may be needed to determine cause or identity.
Can StreamVLM confirm camera obstruction?
No. It reports a match to the visible condition in the prompt. An operator reviews the frame and local context before deciding what it means.
How should a team tune the prompt?
Start with representative cameras, define what makes a useful alert, then adjust the camera selection, confidence threshold, and alert cooldown during a camera-specific pilot.
Is there a published accuracy figure for this use case?
IREX does not publish a universal accuracy figure for this prompt. Image quality, camera angle, lighting, occlusion, and the local definition of a useful alert all affect results.
Protect the Views Your Team Relies On
Pilot on a small set of high-priority cameras and agree what visual change should create a service or security review.