Industrial Optimization

Introducing Visual Inspection AI for discrete manufacturing processes

Top-ranked computer vision and machine learning capabilities of Google to detect the smallest defects at an unprecedented accuracy and scale

Key Benefits of Visual Inspection AI

Purpose-built for production scale

In production trials, customers improved accuracy by up to
10 times than general purpose ML platform

Leading computer vision & AI

Unlike competing solutions that use simple anomaly detection: detect, classify, and precisely locate multiple defect types in a single image

Start quickly

Build accurate models with up to 300 times fewer human labelled images than general purpose ML platform


Integrate your production quality data with your broader industrial data in Google Cloud with our edge-to-cloud data solutions

Our objective

Our goal is to work together to help

⬤ Build a best-in-class visual inspection solution for manufacturing quality control

⬤ Minimize the labeled data required, powered by unsupervised learning and transfer learning

⬤ Set up and enable production workload

⬤ Scale up or down easily with Google Cloud flexible infrastructure to optimize cost

Visual Inspection AI supports a variety of use cases across many industries


Press shop inspection (scratch, dents, cracks, staining)

Foundry engine block inspection (cracks, deformation, anomaly)

Body shop welding seam inspection

Paint shop surface inspection


Wafer level anomaly and defect localization

Die crack inspection

Pre-place inspection

SoC packaging inspection

Board assembly inspection


PCB soldering and gluing (insufficient solder, Icicle, shift, exceeding tin, etc.)

PCB missing components (screw, spring, foam, connector, shield, etc.)

Product surface check (glue spill, mesh deformation, scratches, bubbles, etc.)


Packaging and label inspection

Fabrics inspection (mesh, tear, yarn)

Metal and plastic welding seam inspection

Surface inspection

Visual Inspection makes it easy for manufacturers to extract value from AI

High accuracy

Train high quality defect detection model optimized for inspection use cases

Less training data required

Minimize the labeled data required for model training, powered by unsupervised learning, and transfer learning

On-premises support

Run inference on prem to provide low latency

Quick setup

Set up and enable production workload within several weeks

High scalability

Scale up or down easily with Google Cloud’s flexible infrastructure to optimize cost

Integration with Google Cloud products

Leverage solutions like Anthos, and ML Edge for factory automation

Solution Partners

Contact Sales