Python Computer Vision Project
Real-Time Visual Inspection Using YOLO and OpenCV
A real-time visual inspection system for tracking mayonnaise containers, detecting visible defects, classifying inspection results, and generating reports from camera or video input.
This project explores an automated visual-inspection workflow for mayonnaise containers. It processes camera or video input, tracks individual containers through an inspection region, evaluates visible defect signals, records the inspection result, and prepares data for future custom-model training.
Designed and developed the Python inspection pipeline, camera and video processing, object-tracking workflow, inspection reporting, failed-item snapshot capture, fallback logic, and dataset-collection utilities.
The project is currently a prototype and dataset-collection baseline. It is not presented as a final industrial quality-control product or a publicly deployed factory system.
Camera or video frames enter the inspection pipeline, where containers can be detected and tracked with sequential identifiers. Each tracked item passes through a configured inspection region or gate, receives a PASS or FAIL result, and can produce a CSV record and a failed-item snapshot.
Verified scope
Configured targets
Implementation
Inspection outcomes can be written to CSV reports, failed inspections can trigger automatic snapshots, and optional recording can preserve video output. Dataset utilities support image collection, YOLO-ready preparation, label validation, and dataset auditing.
The current version is a calibration, inspection, and dataset-collection baseline. Production-quality defect detection requires a trained custom YOLO model and validation using representative factory data.
Exploring a computer-vision workflow?