fcddAnomalyDetector

R2026b

Detect anomalies using fully convolutional data description (FCDD) network

Since R2022b

Description

The fcddAnomalyDetector object detects images of anomalies using a fully convolutional data description (FCDD) network. To detect anomalous images, pass the trained detector to the classify function.

Note

This functionality requires Deep Learning Toolbox™.

Creation

Syntax

detector = fcddAnomalyDetector(network)

Description

detector = fcddAnomalyDetector(network) creates an FCDD anomaly detector from a base feature extraction backbone network.

Input Arguments

network — Network

dlnetwork object

Network that serves as the base feature extraction backbone, specified as a dlnetwork (Deep Learning Toolbox) object. You can create a dlnetwork for several common named pretrained networks, such as GoogLeNet and ResNet-18, using the pretrainedEncoderNetwork function.

Properties

Threshold — Anomaly threshold

[] (default) | numeric scalar

Anomaly threshold, specified as a numeric scalar. You must set the value of the threshold before calling the classify function. You can calculate a threshold using the anomalyThreshold function.

Layers — Network layers

Read-only: array of Layer objects

This property is read-only.

Network layers, specified as an array of Layer objects.

InputNames — Network input layer names

Read-only: [] (default) | cell array of character vectors

This property is read-only.

Network input layer names, specified as a cell array of character vectors.

OutputNames — Network output layer names

Read-only: [] (default) | cell array of character vectors

This property is read-only.

Network output layer names, specified as a cell array of character vectors.

State — Network state

Read-only: table

This property is read-only.

Network state, specified as a table with three columns:

• Layer – Layer name, specified as a string scalar.

• Parameter – State parameter name, specified as a string scalar.

• Value – Value of state parameter, specified as a dlarray object.

Layer states contain information calculated during the layer operation to be retained for use in subsequent forward passes of the layer. For example, the cell state and hidden state of LSTM layers, or running statistics in batch normalization layers.

Learnables — Network learnable parameters

Read-only: table

This property is read-only.

Network learnable parameters, specified as a table with three columns:

• Layer – Layer name, specified as a string scalar.

• Parameter – Parameter name, specified as a string scalar.

• Value – Value of parameter, specified as a dlarray object.

The network learnable parameters contain the features learned by the network. For example, the weights of convolution and fully connected layers.

Object Functions

predict	Predict unnormalized anomaly scores
classify	Classify image as normal or anomalous
anomalyMap	Predict per-pixel anomaly score map

Examples

Create FCDD Anomaly Detector

Create an fcddAnomalyDetector object with a ResNet-18 backbone.

backbone = pretrainedEncoderNetwork("resnet18",3);
detector = fcddAnomalyDetector(backbone)

detector = 
  fcddAnomalyDetector with properties:

      Threshold: []
         Layers: [43×1 nnet.cnn.layer.Layer]
     InputNames: {'data'}
    OutputNames: {'fcddHeadPseudoHuber'}
          State: [4×3 table]
     Learnables: [10×3 table]

References

[1] Liznerski, Philipp, Lukas Ruff, Robert A. Vandermeulen, Billy Joe Franks, Marius Kloft, and Klaus-Robert Müller. "Explainable Deep One-Class Classification." Preprint, submitted March 18, 2021. https://arxiv.org/abs/2007.01760.

Extended Capabilities

C/C++ Code Generation

Generate C and C++ code using MATLAB® Coder™.

Usage notes and limitations:

• To create an fcddAnomalyDetector object for code generation, use the coder.loadDeepLearningNetwork (MATLAB Coder) function. For information on how to load the detector for code generation, see Load Pretrained Networks for Code Generation (MATLAB Coder).

GPU Code Generation

Generate CUDA® code for NVIDIA® GPUs using GPU Coder™.

Refer to the usage notes and limitations in the C/C++ Code Generation section. The same usage notes and limitations apply to GPU code generation.

Version History

Introduced in R2022b

R2026b: Moved to Visual Inspection Toolbox

Starting in R2026b, fcddAnomalyDetector is available in the Visual Inspection Toolbox™ product. In releases R2026a and before, fcddAnomalyDetector is available in the Automated Visual Inspection Library for Computer Vision Toolbox™ add-on.

R2024b: Use coder.loadDeepLearningNetwork for code generation

Starting in R2024b, you can use the coder.loadDeepLearningNetwork function to load a fcddAnomalyDetector object for CPU and GPU code generation.

R2023a: Support for C and GPU code generation

fcddAnomalyDetector now supports the generation of C code (requires MATLAB^® Coder™) and optimized CUDA^® code (requires GPU Coder™).