One operating surface
Move from model inference to a field-operable application.
The console exposes system resources, running workloads, live analysis, alarms, and downstream delivery without splitting the project into disconnected demos.
Official product demos
See the complete
edge AI workflow.
These stable, first-party playback pages are the canonical video destination for GitHub, Gitee, documentation, and product links.
Product overview
Operate video AI as one edge system.
See the console bring runtime status, scene tasks, live results, and event delivery together in one operating surface.
Visual orchestration
Compose a complete video analytics pipeline.
Connect decoding, inference, tracking, rules, overlays, and reporting as a reusable task graph.
Reusable task graphs
Turn project logic into a workflow that can be reviewed and replicated.
The visual pipeline makes the executed stages explicit while preserving target-specific models and runtime boundaries.
Prompt-driven vision
Bring DINO and VLM judgment into edge workflows.
Describe open-vocabulary targets or visual states, then connect the result to regions, rules, screenshots, and event outputs.
Long-tail scenarios
Validate a visual task before committing to another training cycle.
Prompt-based DINO and VLM stages can work alongside conventional CV models while keeping the final event reviewable.
Build the first workflow
Continue from the demo into a reproducible setup.
Use the official tutorials for installation, model preparation, task configuration, and platform-specific validation boundaries.