Smart Defect Inspection Application

AI powered solution designed to reduce overkill, improve yield and visual inspector’s health

Hi-precision industries like semiconductor manufacturing demand near-zero defects in its assembly line for curtailing cost to quality. However, existing methods of defect inspection are prone to human errors & inefficiency. Visual inspection or semi-automated inspection methods often end up passing defective components, leading to under-utilization of expensive equipment and adversely affecting the visual inspector’s health.

QuEST’s Smart Defect Inspection (SDI) application is built to resolve the most pressing challenges faced by manufacturing industries. SDI application is a GUI based defect inspection service powered by Artificial Intelligence (AI). QuEST’s SDI can enable real-time defect inspection & diagnosis and ensures optimal equipment usage, product quality inspections, and prognosis.

Objectives of QuEST's SDI Application

Potential Impact

Reduce overkill by up to 30% which in turn improves manufacturing yield

Achieved a mean IoU (Intersection over Union) metric of 0.9

Significant Improvement in Overall Equipment Effectiveness (OEE)

Increased inspector’s break time

Architecture

The SDI application is an AI powered solution for automation of industrial defect inspection, which is capable of localizing small defect areas apart from classification and detection. Its deep learning network is based on trained U-NET architecture to segment defects with inference on the EDGE. The application supports 3 modes of execution: Single mode, Batch mode and Cine mode. A domain expert can control the rate of overkill and underkill using slider (for threshold) on inspection of defect size/area and proximity to design features.
The SDI Application's Architecture

Successful Implementation

The SDI POC has been successfully implemented with two leading semiconductor clients of QuEST. The first implementation detects defects from Scanning Electron Microscope (SEM) & optical inspection images, while the second implementation classifies semiconductor wafer defects.

Smart Defect Inspection Application Overview

Play Video

SDI Application​ Demo

Play Video

Contact Us