Hearthlight by Hearthlight provides local-first computer vision automation, enabling users to connect cameras, models, triggers, and actions to transform video into structured events and workflows.
Security monitoring
Use RTSP cameras, HTTP feeds, or local webcams to detect anomalies and trigger alerts without cloud dependency.
Industrial automation
Connect physical systems and automations to respond to visual events in real time.
Retail analytics
Track foot traffic, behaviors, and incidents using local models and structured event outputs.
Research and prototyping
Experiment with computer vision models, triggers, and connectors in a plugin-based runtime environment.
Smart home integration
Run anomaly detection and action workflows from local video sources.
Event logging
Persist runs, incidents, entities, and recordings for follow-up analysis.
Any video source
Process RTSP cameras, HTTP feeds, local webcams, and uploaded video through a single operator workflow.
Scalable YAML configuration
Save default model bindings, source overrides, anomaly prompts, and alert rules without editing raw YAML in the browser.
Outputs trigger anything
Persist runs, incidents, entities, anomaly events, and recordings so follow-up work starts from structured evidence.
Extensible plugin framework
Extend models, triggers, connectors, and optional integrations through restart-loaded plugin catalogs.
Model Zoo
Access detection, tracking, and reasoning AI models that define what the runtime can see.
Trigger Zoo
Define rules, prompts, behaviors, events, thresholds, and workflows that decide when something matters.
Connector Zoo
Connect notifications, APIs, automations, physical systems, and external tools that decide where alerts or actions go.
Local-first runtime
The entire system runs locally with the control plane designed for local operator control.
Open-source core
The runtime and plugin catalogs are open-source, allowing full transparency and customization.
pip install hear. Run hearthlight onboard to initialize the workspace, then inspect the current inventory. Launch the stack with hearthlight start --interactive to open the dashboard. From there, configure cameras, select models from the Model Zoo, set triggers in the Trigger Zoo, and define connectors in the Connector Zoo—all through the browser-based YAML interface.Hearthlight by Hearthlight provides local-first computer vision automation, enabling users to connect cameras, models, triggers, and actions to transform video into structured events and workflows.
Category:Automation
Visit Link:https://hearthlight.net/
Tags:computer vision、automation、local-first、video analysis、workflow