Computer vision answers · Manufacturing

How do I use zones to monitor multiple conveyors from one camera with computer vision?

Without putting a camera, or a person, on every belt.

Short answer

Train an object detection model to find the items on your belts with Roboflow for free, then draw one zone over each conveyor and count the detections inside it. One camera gives you a separate count for every conveyor, so you can see which line is busy, empty or backed up. Draw each zone tightly along its own belt and count items by their bottom center, so items on a neighboring conveyor never land in the wrong zone.

Start free
Roboflow agent · free to start

Skip the setup. Describe your images.

Describe your images and what you need to find. The agent sets up labeling, training and deployment with you.

Key facts

What you need
One camera that sees every conveyor, and a model for your items
What you get
A live item count per conveyor, plus crossings and time in zone
Image sources
IP and RTSP cameras, webcams and recorded video
Runs on
Cloud API, or an edge device or server running Docker
Three roller conveyors in a warehouse, two carrying cardboard boxes, overlaid with digital labels 'Conveyor 01', '02', '03' and live monitoring data.

How to build conveyor zone monitoring, step by step

Six steps from one camera view to a live count for every conveyor.

  1. Connect Roboflow to your coding agent, ChatGPT or Codex

    Add the Roboflow MCP server to Claude, Cursor or Gemini CLI, or install the Roboflow plugin in ChatGPT or Codex. Your assistant can then upload and auto-label camera frames, start training, build the Workflow and run the model from a prompt, at no extra cost beyond your Roboflow plan.

  2. Point one camera at every conveyor

    Mount one camera high enough that every conveyor is in view and the belts do not overlap in the image. Collect frames from that camera at quiet and busy times, in the lighting you will run in.

  3. Label the items on the belts

    Draw a box around every item on the belts, such as parcels, totes or products, and give each item type its own class. Roboflow Annotate can pre-label frames with SAM 3, so you correct boxes instead of drawing each one.

  4. Train a detection model

    Train an object detection model such as RF-DETR on your labeled frames with Roboflow Train.

  5. Draw one zone per conveyor and count what is inside

    Draw a polygon zone over each conveyor in the camera's view, run the model on a frame, and count the detections inside each zone. A conveyor that reads zero while the others are busy is empty.

    Python
    import numpy as np
import supervision as sv
from inference import get_model

model = get_model("your-project/1")
zones = {
    "conveyor_1": sv.PolygonZone(np.array([[50, 300], [1230, 300], [1230, 420], [50, 420]])),
    "conveyor_2": sv.PolygonZone(np.array([[50, 500], [1230, 500], [1230, 620], [50, 620]])),
}

result = model.infer("frame.jpg", confidence=0.4)[0]
detections = sv.Detections.from_inference(result)
for name, zone in zones.items():
    zone.trigger(detections)
    print(name, zone.current_count)
# conveyor_1 6
# conveyor_2 0

    Rather not write code? Ask the Roboflow agent to set this up for you

  6. Track items and measure each conveyor over time

    Build a Workflow that tracks each item across frames, adds one Line Counter per conveyor to count the items that pass, and uses Time in Zone to show items that stay in a zone too long. Run it on your camera stream with Roboflow Deploy in the cloud or on an edge device.

Why 2 million engineers build on Roboflow

  • Label faster

    Auto-label with SAM 3 or your own model, then fix the misses instead of drawing every box.

  • RF-DETR

    Roboflow’s open-source real-time detection model, built for accuracy (opens in a new tab) on small objects.

  • Logic built in

    Counting, zones, tracking and alerts are ready-made Workflow blocks, so there is less code to write.

  • Your model, your weights

    Use your trained weights on every plan, including free, through the open-source Inference package.

  • Run it anywhere

    Self-host on a workstation, Docker, an on-prem GPU or an edge device, or call the cloud API.