Détail de l'auteur
Auteur Abdollah Amirkhani |
Documents disponibles écrits par cet auteur (1)
![](./images/expand_all.gif)
![](./images/collapse_all.gif)
![Tris disponibles](./images/orderby_az.gif)
Adversarial defenses for object detectors based on Gabor convolutional layers / Abdollah Amirkhani in The Visual Computer, vol 38 n° 6 (June 2022)
![]()
[article]
Titre : Adversarial defenses for object detectors based on Gabor convolutional layers Type de document : Article/Communication Auteurs : Abdollah Amirkhani, Auteur ; Mohammad Karimi, Auteur Année de publication : 2022 Article en page(s) : pp 1929 - 1944 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique
[Termes IGN] apprentissage profond
[Termes IGN] détection d'objet
[Termes IGN] filtre de Gabor
[Termes IGN] reconnaissance de formes
[Termes IGN] réseau antagoniste génératifRésumé : (auteur) Despite their many advantages and positive features, the deep neural networks are extremely vulnerable against adversarial attacks. This drawback has substantially reduced the adversarial accuracy of the visual object detectors. To make these object detectors robust to adversarial attacks, a new Gabor filter-based method has been proposed in this paper. This method has then been applied on the YOLOv3 with different backbones, the SSD with different input sizes and on the FRCNN; and thus, six robust object detector models have been presented. In order to evaluate the efficacy of the models, they have been subjected to adversarial training via three types of targeted attacks (TOG-fabrication, TOG-vanishing, and TOG-mislabeling) and three types of untargeted random attacks (DAG, RAP, and UEA). The best average accuracy (49.6%) was achieved by the YOLOv3-d model, and for the PASCAL VOC dataset. This is far superior to the best performance and accuracy and obtained in previous works (25.4%). Empirical results show that, while the presented approach improves the adversarial accuracy of the object detector models, it does not affect the performance of these models on clean data. Numéro de notice : A2022-382 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article DOI : 10.1007/s00371-021-02256-6 Date de publication en ligne : 24/07/2021 En ligne : https://doi.org/10.1007/s00371-021-02256-6 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=100651
in The Visual Computer > vol 38 n° 6 (June 2022) . - pp 1929 - 1944[article]