Computer vision
Real-time GPU pipelines: licence plates through extreme degradation, FaceID with persistent tracking, fire and smoke detection.
FireSmokeDetector
Computer vision fire and smoke detection.
Research Computer vision
A real-time three-tier detection pipeline using YOLO, CLIP and depth estimation, with context-aware hazard scoring, spread prediction and annotated live overlays.
What it does
Built for environments where false positives are expensive and missed detections are catastrophic, so the three tiers disagree deliberately: a fast detector proposes, a vision-language model confirms context, and depth estimation grounds the scale.
Who it is for
Industrial and land-management operators monitoring for fire and smoke on camera.
What it costs
Research project; contact us about deployment.
Terminology
Words this page uses in a specific sense.
- Three-tier pipeline
- Detection, semantic confirmation and geometric grounding as separate stages, so a single-model false positive has to survive two independent checks.
- Spread prediction
- Estimating where a detected fire front is likely to move next from its observed growth and the scene geometry.
Facts
| Field | Value |
|---|---|
| Name | FireSmokeDetector |
| Status | Research |
| Category | Computer vision |
| Platforms | Linux, Windows |
| Pricing | Research project; contact us about deployment. |
| Source | Closed source |
| Publisher | Cognitive Industries |
| Published from | Brisbane, Queensland, Australia |
Capabilities behind it
Related
CogVisIntelligence
Real-time GPU vision: plates and faces.
ChazzAI
AI roleplay and living worlds.
Unslop
Find and fix known bugs. No model in the loop.
Last reviewed · Site changelog