An automated computer vision system designed for industrial manufacturing and packaging environments (such as PepsiCo operations) to perform real-time quality control on product labels using custom YOLOv8 instance segmentation and OpenCV.
Ensuring label integrity on production lines is critical. Defective labels, unreadable barcodes, or misaligned placements can cause supply chain issues and rejection downstream. This system processes high-speed video feeds to monitor and classify labels in real-time, distinguishing between compliant ("good") and non-compliant ("bad") items based on strict visual and positional criteria.
- 🟢 Good Label (
etiqueta buena): The barcode is fully readable, intact, and correctly positioned inside the target validation zone. - 🔴 Bad Label (
etiqueta mala): Triggered when the barcode is illegible, torn, poorly printed, or positioned outside the designated bounding box area.
- Python 3.14
- Ultralytics YOLOv8 (Custom Instance Segmentation)
- OpenCV (
cv2) (Video processing, masking, and real-time rendering) - Cvzone (Visual UI overlays and text bounding)
- NumPy (Data and coordinate manipulation)
etiquetas/
│
├── yolov8-custom-segment-main/
│ ├── main1.py # Main execution script for video processing
│ ├── yolo_segmentation.py # YOLOv8 custom wrapper class for segmentation masks
│ ├── best.pt # Trained YOLOv8 custom weights model
│ ├── coco1.txt # Class labels configuration file
│ └── etiquetas.mp4 # Sample input video feed for testing
│
└── README.md