Menu
Deploy a trained YOLO model as a Nuclio serverless function for CVAT auto-annotation.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
| name | cvat-deploy |
| description | Deploy a trained YOLO model as a Nuclio serverless function for CVAT auto-annotation. |
Deploy a trained YOLO model as a Nuclio serverless function so CVAT users can auto-annotate images directly from the CVAT UI.
yolo-project.yaml has class definitionsexperiments/summary.md for the best-performing model pathyolo-cvat deploy --model <path_to_model> --name <detector_name>
This generates:
serverless/<name>/function.yaml โ Nuclio config with class specserverless/<name>/main.py โ Inference handlerserverless/<name>/best.onnx โ Exported modelRead the generated function.yaml and confirm:
yolo-project.yamlProvide the command for the user to run:
nuctl deploy --path ./serverless/<name> --platform local
nuctl get functions
Check that the function is running.
Model is .pt format?
โโโ Yes โ Export to ONNX first (yolo-export)
โโโ Already .onnx โ Copy directly
Nuclio reachable on port 8070?
โโโ Yes โ Ready to deploy
โโโ No โ Warn user, suggest checking Docker and CVAT stack
Function with same name exists?
โโโ Yes โ Ask user: overwrite or use different name?
โโโ No โ Deploy normally
Set up autonomous training monitoring โ creates cron jobs to track long-running training, auto-continue pipeline when training completes.
Audit YOLO dataset quality โ class distribution, annotation quality, image stats, and improvement suggestions.
Orchestrate the full active learning loop: train, analyze, push to CVAT, wait for review, pull, merge, retrain.
Analyze YOLO training runs โ compares to baseline/best, checks per-class regression, analyzes training dynamics and tune convergence, writes actionable recommendations.
Run autonomous YOLO training experiments โ reads training-plan.md, assesses bottlenecks, acts strategically, and delegates HP optimization to model.tune().
Profile YOLO model inference speed, FPS, and size across image sizes and export formats.