Deploying iGromi Cam (AI Vision)

The industrial eye of iGromi

When traditional sensors cannot detect quality defects, colours, or overlapping products on a conveyor, we use iGromi Cam paired with our inference models running at the edge (YOLO/OpenCV).

1. Physical mounting

1. Place the camera perpendicular to the conveyor. The ideal focal distance is between 40cm and 120cm depending on the lens.
2. Fix the articulated arm to the machine chassis using M6 bolts. Make sure machine vibration does not reach the lens directly.
3. Lighting: Computer vision requires constant light. Install an LED ring light or diffuser bars if plant lighting fluctuates.

2. Data and power

The camera uses PoE (Power over Ethernet).

• Run a Cat6 cable from the camera to the iGromi PoE switch in the cabinet.
• Do not exceed 90 metres of cabling.

3. Model calibration (dashboard)

1. Open your iGromi OS Central panel > AI Vision.
2. Select the newly detected camera (it appears as Pending Provisioning).
3. Click Capture Base Frame.
4. Draw the Bounding Box (detection zone) on screen. This is the area where the algorithm will count products or look for defects.
5. Set the Confidence Threshold to 85% to avoid false positives.

The model is compiled and pushed to the Edge Node on the machine over the air (OTA). Within 60 seconds it starts streaming MQTT telemetry.