← Projects

AI / ML

Applying Computer Vision to Smart Manufacturing Quality Control

A smart manufacturing case study focused on defect detection, model comparison, and production practicality.

Role
AI/ML engineering with focus on computer-vision experimentation and production constraints.
Context
WiMetrix
PythonComputer VisionSSDFaster R-CNNYOLOv3YOLOv4
Architecture Concept

A smart manufacturing case study focused on defect detection, model comparison, and production practicality.

Production Imagery
Preprocessing
CV Model
Defect Detection
Production Analytics

Overview

WiMetrix is an AI and RFID-powered smart-factory solution aimed at textile and apparel manufacturing.

My role included AI/ML engineering related to computer vision for detecting stitching and manufacturing defects.

Business Context

Manufacturing platforms need visibility into garment movement, workforce and production efficiency, quality, work in progress, bottlenecks, and production analytics.

Technical Work

The work involved evaluating approaches such as SSD, the R-CNN family, Faster R-CNN, YOLOv3, and YOLOv4.

The engineering theme was not simply model accuracy. It was accuracy versus inference speed, operational constraints, and production practicality.

System Architecture

The conceptual pipeline moved from production imagery to preprocessing, computer-vision inference, defect detection, quality-control information, and production analytics.

Trade-offs

More complex detection approaches may improve quality in some contexts but can introduce heavier inference costs. Production environments require a practical balance between detection quality, speed, deployment complexity, and maintainability.

Outcome

The work supported computer-vision experimentation and integration thinking for manufacturing quality workflows without publishing proprietary datasets, accuracy figures, or model internals.