Logo del repository
  1. Home
 
Opzioni

Masked Transformer for Image Anomaly Localization

De Nardin, Axel
•
Mishra, Pankaj
•
Foresti, Gian Luca
•
Piciarelli, Claudio
2022
  • journal article

Periodico
INTERNATIONAL JOURNAL OF NEURAL SYSTEMS
Abstract
Image anomaly detection consists in detecting images or image portions that are visually different from the majority of the samples in a dataset. The task is of practical importance for various real-life applications like biomedical image analysis, visual inspection in industrial production, banking, traffic management, etc. Most of the current deep learning approaches rely on image reconstruction: the input image is projected in some latent space and then reconstructed, assuming that the network (mostly trained on normal data) will not be able to reconstruct the anomalous portions. However, this assumption does not always hold. We thus propose a new model based on the Vision Transformer architecture with patch masking: the input image is split in several patches, and each patch is reconstructed only from the surrounding data, thus ignoring the potentially anomalous information contained in the patch itself. We then show that multi-resolution patches and their collective embeddings provide a large improvement in the model's performance compared to the exclusive use of the traditional square patches. The proposed model has been tested on popular anomaly detection datasets such as MVTec and head CT and achieved good results when compared to other state-of-the-art approaches.
DOI
10.1142/S0129065722500307
WOS
WOS:000823075300001
Archivio
http://hdl.handle.net/11390/1230667
info:eu-repo/semantics/altIdentifier/scopus/2-s2.0-85133005282
https://www.worldscientific.com/doi/10.1142/S0129065722500307
https://ricerca.unityfvg.it/handle/11390/1230667
Diritti
metadata only access
Soggetti
  • Anomaly detection

  • image inpainting

  • self-supervised learn...

  • vision transformer

  • Image Processing, Com...

  • Tomography, X-Ray Com...

google-scholar
Get Involved!
  • Source Code
  • Documentation
  • Slack Channel
Make it your own

DSpace-CRIS can be extensively configured to meet your needs. Decide which information need to be collected and available with fine-grained security. Start updating the theme to match your nstitution's web identity.

Need professional help?

The original creators of DSpace-CRIS at 4Science can take your project to the next level, get in touch!

Realizzato con Software DSpace-CRIS - Estensione mantenuta e ottimizzata da 4Science

  • Impostazioni dei cookie
  • Informativa sulla privacy
  • Accordo con l'utente finale
  • Invia il tuo Feedback