Automated Visual Inspection

Automate quality assurance tasks using anomaly detection and classification techniques

Automated visual inspection (AVI) is a set of techniques used to determine whether an image represents a normal ("good") state or an anomalous ("defective") state. AVI assists and improves quality assurance processes commonly found in manufacturing settings. Modern visual inspection uses machine learning and deep learning techniques to produce useful results.

The specific technique you select to automate a visual inspection task depends on several factors. These factors include the amount of training data available for normal and anomalous samples, the number of anomaly classes to recognize, and the type of localization information required for understanding and monitoring predictions.

To perform automated visual inspection, download the Computer Vision Toolbox™ Automated Visual Inspection Library from the Add-On Explorer. For more information on downloading add-ons, see Get and Manage Add-Ons. Some functionality also requires Deep Learning Toolbox™.

Functions

expand all

Load Training Data

`groundTruth`	Ground truth label data
`imageDatastore`	Datastore for image data
`sceneLabelTrainingData`	Create training data for scene classification from ground truth
`splitAnomalyData`	Split data into training, validation and testing sets for anomaly detection

Train Anomaly Detector

`trainFCDDAnomalyDetector`	Train fully convolutional data description (FCDD) anomaly detection network
`trainFastFlowAnomalyDetector`	Train FastFlow anomaly detection network
`trainPatchCoreAnomalyDetector`	Train PatchCore anomaly detection network
`anomalyThreshold`	Optimal anomaly threshold for set of anomaly scores and corresponding labels

Detect Anomalies Using Deep Learning

`fcddAnomalyDetector`	Detect anomalies using fully convolutional data description (FCDD) network for anomaly detection
`fastFlowAnomalyDetector`	Detect anomalies using FastFlow network
`patchCoreAnomalyDetector`	Detect anomalies using PatchCore network
`classify`	Classify image as normal or anomalous
`predict`	Predict unnormalized anomaly scores

Visualize and Evaluate Results

`anomalyMap`	Predict per-pixel anomaly score map
`anomalyMapOverlay`	Overlay heatmap on image using per-pixel anomaly scores
`viewAnomalyDetectionResults`	View anomaly detection results
`evaluateAnomalyDetection`	Evaluate anomaly detection results against ground truth
`anomalyDetectionMetrics`	Anomaly detection metrics

Topics

Getting Started with Anomaly Detection Using Deep Learning
Anomaly detection using deep learning is an increasingly popular approach to automating visual inspection tasks.

Featured Examples

Detect Defects on Printed Circuit Boards Using YOLO v4 Network

Detect, localize, and classify defects in printed circuit boards (PCBs) using a you only look once version 4 (YOLO v4) deep learning network.

Open Live Script

Classify Defects on Wafer Maps Using Deep Learning

Classify manufacturing defects on wafer maps using a simple convolutional neural network (CNN).

Open Live Script

Detect Image Anomalies Using Explainable FCDD Network

Use an anomaly detector to distinguish between normal pills and pills with anomalous chips or contamination.

Open Live Script

Detect Anomalies in Pills During Live Image Acquisition

Detect anomalies in pills during live image acquisition.

Detect Image Anomalies Using Pretrained ResNet-18 Feature Embeddings

Train a similarity-based anomaly detector using one-class learning of feature embeddings extracted from a pretrained ResNet-18 convolutional neural network.

Open Live Script