Home / AI model / Live fire test
Browser demonstration

AI flame detection in action

Upload a photograph and the Detectium model will analyse it for visible flames. Review the marked image, confidence score and a concise machine-readable result on one page.

Choose an image How it works

Demonstration only. A result is not a fire alarm or a substitute for certified fire-safety systems and procedures.

Two visible flames identified with confidence boxes in an industrial facility
Visual AI analysis · visible-flame detection
Live model test

Test a fire image

Use the sample image or choose a JPEG, PNG or WebP image up to 8 MB. Images are sent to the Detectium detection service for analysis and are not presented as training data.

01

Input image

Ready
Image selected for fire analysis
Drop an image hereor choose a file below

Sample image selected

02

Detection result

Waiting for image
RUN MODEL

Your marked image will appear here.

Fire statusNot analysed
Flames found—
Top confidence—
View machine-readable result 
Run the model to view the response.
InputImageJPEG, PNG or WebP
Current targetFlameVisible-fire detection only
OutputMarkedImage, count and confidence
IntegrationJSONStructured detection result
Model and integration

Built for rapid fire analysis

The live test brings the model, its structured output and developer references together in one place.

Lightning Fast

Optimized computer vision supports rapid image analysis for demonstrations and integration testing.

Developer Ready

The same backend exposes model discovery, structured JSON predictions and server-annotated JPEG output.

Highly Accurate

Trained across varied visual environments to support reliable visible-flame detection while reducing false positives.

DEVELOPER RESOURCES

Explore the detection API

Review all three verified endpoints in the readable guide, interactive Swagger interface or ReDoc reference.

API documentation → Open Swagger → Open ReDoc →
How it works

From image to operator-readable evidence

The demonstration reflects the core visual-analysis flow while keeping the result focused on visible flames.

01 · INGEST

Prepare the image

The browser validates the image format and sends it securely to the detection service.

JPEG / PNG / WEBP · MAX 8 MB
02 · ANALYSE

Find visible flame signatures

The computer-vision model evaluates the scene and returns candidate flame regions with confidence values.

CURRENT CAPABILITY · VISIBLE FLAME
03 · REVIEW

Inspect the result

The page draws the returned fire regions over the source image and presents a concise structured summary.

IMAGE OVERLAY + JSON SUMMARY
Deployment planning

Want to test Detectium on-device AI solution at your facility?

Share your building, operating conditions and detection requirements with our team.

Discuss a pilot →