Automated Food-Packaging Layout Inspection

Overview

Built an automated visual inspection system that verifies food-packaging layout on a production conveyor line, checking each tray’s contents against an expected layout as it moves through the line.

Technical Architecture

graph LR
    classDef trigger fill:#e1f5fe,stroke:#01579b,stroke-width:2px,color:#000000;
    classDef process fill:#f3e5f5,stroke:#4a148c,stroke-width:2px,color:#000000;
    classDef gate fill:#fff3e0,stroke:#e65100,stroke-width:2px,color:#000000;
    classDef result fill:#e8f5e9,stroke:#1b5e20,stroke-width:2px,color:#000000;

    A(["Camera Frame"]) --> B["Presence Gate<br/>(skip empty frames)"]
    B --> C["Detect Tray & Items<br/>(open-vocabulary model)"]
    C --> D["Rectify to<br/>Top-Down View"]
    D --> E["Compare Slots vs.<br/>Expected Layout"]
    E --> F{"Confirmed Over<br/>N Frames?"}
    F -->|Correct| G["Track Resolved"]
    F -->|Mismatch| H["Alert + Log"]

    class A trigger;
    class B,C,D,E process;
    class F gate;
    class G,H result;

Technical Approach

• Open-Vocabulary Detection: Uses a promptable segmentation model rather than a fixed-class detector, so tray and item categories can be redefined without retraining
• Perspective-Normalized Verification: Detected trays are rectified to a top-down view via a perspective transform, then each item is assigned to a slot and checked against an expected layout — a config-driven mapping rather than pixel-level template matching
• Adaptive Compute for Line Speed: A lightweight presence check filters out empty frames before running the heavier detection model, with inference rate scaled up only when a tray is actually on the line
• Multi-Frame Confirmation: A verdict must hold for several consecutive frames as a tray crosses the inspection zone before it’s finalized, cutting false positives from single-frame noise

Key Features

• Real-Time Quality Control: Flags layout mismatches before products leave the line
• Config-Driven Layout Rules: Expected item placement is defined as data, not code, so new tray layouts don’t require redeploying the detection logic
• Audit Trail: Every flagged tray is logged with annotated images for human review

Technologies Used