Towards Total Recall in Industrial Anomaly Detection (2021-06-15T00:00:00.000000Z)

TL;DR

This work proposes PatchCore, which uses a maximally representative memory bank of nominal patch-features, which offers competitive inference times while achieving state-of-the-art performance for both detection and localization.

Abstract

Being able to spot defective parts is a critical component in large-scale industrial manufacturing. A particular challenge that we address in this work is the cold-start problem: fit a model using nominal (non-defective) example images only. While handcrafted solutions per class are possible, the goal is to build systems that work well simultaneously on many different tasks automatically. The best peforming approaches combine embeddings from ImageNet models with an outlier detection model. In this paper, we extend on this line of work and propose PatchCore, which uses a maximally representative memory bank of nominal patch-features. PatchCore offers competitive inference times while achieving state-of-the-art performance for both detection and localization. On the challenging, widely used MVTec AD benchmark PatchCore achieves an image-level anomaly detection AUROC score of up to 99.6%, more than halving the error compared to the next best competitor. We further report competitive results on two additional datasets and also find competitive results in the few samples regime. Code: github.com/amazon-research/patchcore-inspection.

Authors

B. Scholkopf

18 papers

T. Brox

22 papers

Peter Gehler

4 papers

TL;DR

Abstract

Authors

References61 items

Multiresolution Knowledge Distillation for Anomaly Detection

PaDiM: a Patch Distribution Modeling Framework for Anomaly Detection and Localization

Same Same But DifferNet: Semi-Supervised Defect Detection with Normalizing Flows

Patch SVDD: Patch-level SVDD for Anomaly Detection and Segmentation

Modeling the Distribution of Normal Data in Pre-Trained Deep Features for Anomaly Detection

Sub-Image Anomaly Detection with Deep Pyramid Correspondences

RaPP: Novelty Detection with Reconstruction along Projection Pathway

Deep Nearest Neighbor Anomaly Detection

Revisiting Training Strategies and Generalization Performance in Deep Metric Learning

Iterative energy-based projection on a normal data manifold for anomaly localization

Unsupervised Anomaly Detection for X-Ray Images

PyTorch: An Imperative Style, High-Performance Deep Learning Library

Inverse-Transform AutoEncoder for Anomaly Detection

Attention Guided Anomaly Localization in Images

Towards Visually Explaining Variational Autoencoders

Uninformed Students: Student-Teacher Anomaly Detection With Discriminative Latent Embeddings

Small-GAN: Speeding Up GAN Training Using Core-sets

Data-Independent Neural Pruning via Coresets

XRAI: Better Attributions Through Regions

MVTec AD — A Comprehensive Real-World Dataset for Unsupervised Anomaly Detection

Memorizing Normality to Detect Anomaly: Memory-Augmented Deep Autoencoder for Unsupervised Anomaly Detection

OCGAN: One-Class Novelty Detection Using GANs With Constrained Latent Representations

Approximating CNNs with Bag-of-local-Features models works surprisingly well on ImageNet

Are pre-trained CNNs good feature extractors for anomaly detection in surveillance videos?

Anomaly Detection With Multiple-Hypotheses Predictions

Image Anomaly Detection with Generative Adversarial Networks

Framework

Surface defect saliency of magnetic tile

Glow: Generative Flow with Invertible 1x1 Convolutions

Generative Probabilistic Novelty Detection with Adversarial Autoencoders

Improving Unsupervised Defect Segmentation by Applying Structural Similarity to Autoencoders

GANomaly: Semi-Supervised Anomaly Detection via Adversarial Training

Deep Anomaly Detection Using Geometric Transformations

Adversarially Learned One-Class Classifier for Novelty Detection

Deep Autoencoding Gaussian Mixture Model for Unsupervised Anomaly Detection

Future Frame Prediction for Anomaly Detection - A New Baseline

Active Learning for Convolutional Neural Networks: A Core-Set Approach

Billion-Scale Similarity Search with GPUs

Aggregated Residual Transformations for Deep Neural Networks

Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization

Transfer representation-learning for anomaly detection

Density estimation using Real NVP

Deep Structured Energy Based Models for Anomaly Detection

Wide Residual Networks

Deep Residual Learning for Image Recognition

Distilling the Knowledge in a Neural Network

Anomaly Detection Using Autoencoders with Nonlinear Dimensionality Reduction

Scalable Training of Mixture Models via Coresets

Support Vector Data Description

ImageNet: A large-scale hierarchical image database

Python 3 Reference Manual

Coresets, sparse greedy approximation, and the Frank-Wolfe algorithm

Smaller Coresets for k-Median and k-Means Clustering

Image quality assessment: from error visibility to structural similarity

An elementary proof of a theorem of Johnson and Lindenstrauss

Support Vector Method for Novelty Detection

Pytorch image models

Integer and Combinatorial Optimization

Geometric Approximation via Coresets

A Geometric Framework for Unsupervised Anomaly Detection

On the generalized distance in statistics

Field of Study

Journal Information

Name

Page

Venue Information

Name

Type

URL

Alternate Names