Computer-Vision Quality-Control Platform

Overview

A production computer-vision quality-control platform built at Lenskart that automatically extracts 17 critical eyeglass-frame measurements, generates inspection annotations and centralizes the entire quality-review workflow.

It runs on a distributed two-machine architecture — one for image acquisition, one for centralized processing and automated data management — designed to scale inspection operations. A measurement engine built with Python, OpenCV, NumPy and Streamlit turns captured images into annotated, reviewable inspection outputs with full traceability.

Highlights

• Designed a distributed two-machine architecture for acquisition, centralized processing and automated data management.
• Built a measurement engine extracting 17 critical frame dimensions and generating annotated inspection outputs.
• Implemented automated image ingestion, file-watcher pipelines, centralized logging and expert-review dashboards.
• Established a structured inspection dataset to power future defect-detection and quality-classification models.