Supervision Draw

Send an image and detections, get the annotated image back. The drawing is done by Roboflow supervision's own annotators, so there is no Python environment to set up: an alternative to installing supervision just to render detection results.

The input scene The same scene with boxes and labels drawn

Annotators: box, round_box, box_corner, circle, dot, ellipse, triangle, label, color, mask, polygon, halo, background_overlay, blur, pixelate and percentage_bar, applied in the order you list them. Up to 4 megapixels and 500 detections per call; labels are ASCII.

Pricing

Pay per call in USDC with x402, or buy prepaid credits by card.

EndpointWhat it doesPriceFree

Try it

curl -X POST "$ORIGIN/v1/annotate" -H 'content-type: application/json' -H 'Authorization: Bearer <your API key>' -d '{
  "image": "'"$(base64 -w0 scene.jpg)"'",
  "detections": [
    {"xyxy": [30, 60, 90, 120], "class_id": 0, "confidence": 0.94},
    {"xyxy": [140, 40, 180, 115], "class_id": 1, "confidence": 0.88}
  ],
  "class_names": ["car", "person"],
  "annotators": [{"type": "box", "thickness": 2}, "label"],
  "format": "jpeg"
}' | jq -r .image | base64 -d > annotated.jpg

The answer is JSON: image (base64), mimeType, width, height and counts per class. Every call is paid (x402 or credits); an unpaid call gets 402 Payment Required with the x402 terms, which x402 clients and agents pay automatically. mask, polygon and halo need a polygon on every detection.

Use it from an AI agent (MCP)

Every endpoint is also a tool on this product's MCP server. Add it to any MCP client (Claude, Cursor, VS Code and others) with your API key:

{ "mcpServers": { "this-api": { "url": "$ORIGIN/mcp", "headers": { "Authorization": "Bearer " } } } }

Buy credits