Inpainting Transformer for Anomaly Detection

28 Apr 2021  ·  Jonathan Pirnay, Keng Chai ·

Anomaly detection in computer vision is the task of identifying images which deviate from a set of normal images. A common approach is to train deep convolutional autoencoders to inpaint covered parts of an image and compare the output with the original image. By training on anomaly-free samples only, the model is assumed to not being able to reconstruct anomalous regions properly. For anomaly detection by inpainting we suggest it to be beneficial to incorporate information from potentially distant regions. In particular we pose anomaly detection as a patch-inpainting problem and propose to solve it with a purely self-attention based approach discarding convolutions. The proposed Inpainting Transformer (InTra) is trained to inpaint covered patches in a large sequence of image patches, thereby integrating information across large regions of the input image. When training from scratch, in comparison to other methods not using extra training data, InTra achieves results on par with the current state-of-the-art on the MVTec AD dataset for detection and surpassing them on segmentation.

PDF Abstract

Datasets


Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Benchmark
Anomaly Detection AeBAD-S InTra Segmentation AUPRO 74.7 # 4
Detection AUROC 44.2 # 6
Anomaly Detection AeBAD-V InTra Detection AUROC 54.1 # 7
Anomaly Detection MVTec AD InTra Detection AUROC 95.0 # 64
Segmentation AUROC 96.6 # 60

Methods