EVALUATION OF THE ROBUSTNESS OF OBJECT DETECTION MODELS AGAIUST ADVERSARIAL ATTACKS

Samatov, Boburmirzo Khamidullaevich, Yunusov, Odiljon Pozilovich, Sayidova, Nigora Komildjonovna, Саматов, Бобурмирзо Хамидуллаевич, Юнусов, Одилжон Позилович, Саидова, Нигора Комилджоновна, Samatov, Boburmirzo Xamidillo o‘g‘li, Yunusov, Odiljon Pozilovich, Sayidova, Nigora Komildjonovna

Ilim ha’m ja’miyet · 2026-yil

Annotatsiya

In this study, object detection systems based on convolutional neural networks were investigated, particularly the effectiveness of adversarial attacks against the YOLOv5 model. As attack methods, the Fast Gradient Sign Method (FGSM) and its improved variants (Iterative FGSM and PGD) were applied. Adversarial perturbations generated using the ResNet18 model were used to compare the object detection results of the original image and the attacked image. The results demonstrated that even a small amount of noise can significantly mislead an object detection model. Adversarial attacks represent one of the major vulnerabilities of artificial intelligence models, especially deep learning–based image recognition systems. Attacks such as FGSM and PGD modify the model input with small perturbations, forcing the model to produce incorrect classifications.

Maqola ma’lumotlari
MualliflarSamatov, Boburmirzo Khamidullaevich, Yunusov, Odiljon Pozilovich, Sayidova, Nigora Komildjonovna, Саматов, Бобурмирзо Хамидуллаевич, Юнусов, Одилжон Позилович, Саидова, Нигора Комилджоновна, Samatov, Boburmirzo Xamidillo o‘g‘li, Yunusov, Odiljon Pozilovich, Sayidova, Nigora Komildjonovna
JurnalIlim ha’m ja’miyet
Nashr sanasi2026-03-31
Son2-1
Betlar38-40
Tiluz_Latn

Maqolaning asl sahifasi

Kalit so‘zlar

FGSM, YOLOv5, PGD, Faster R-CNN, ResNet18, CNN, FGSM, YOLOv5, PGD, Faster R-CNN, ResNet18, CNN, FGSM, YOLOv5, PGD, Faster R-CNN, ResNet18, CNN

Ilmiy soha

BiologiyaFilologiyaFizika-matematikaGeografiyaPedagogikaPsixologiyaTarixTexnika

Ilim ha’m ja’miyet jurnalidan boshqa maqolalar

Ilim ha’m ja’miyet — barcha maqolalar