Amira Guesmi
Papers
6
Total Citations
87
H-Index
4
About
Amira Guesmi is a leading researcher at the intersection of adversarial machine learning and autonomous systems, specializing in the security and robustness of vision-based deep neural networks (DNNs). Her work critically examines how physical adversarial attacks—such as manipulated raindrops, stealthy patches, and shape-sensitive perturbations—can deceive camera-based smart systems, from autonomous vehicles to intelligent robots. Guesmi’s major contributions include pioneering the “AdvRain” framework, which demonstrates how adversarial raindrops can compromise vision modules, and developing “SAAM” and “SSAP,” advanced attacks that disrupt monocular depth estimation (MDE) in navigation applications. Her comprehensive survey on physical adversarial attacks (45 citations) has become a foundational reference, mapping the field’s trends, challenges, and future directions. With over 87 total citations and a rapidly growing body of work, Guesmi’s research highlights critical vulnerabilities in safety-critical AI, pushing the community toward more resilient perception systems. Her innovative, application-driven approach—blending theoretical rigor with real-world threat modeling—makes her a vital voice in ensuring the reliability of next-generation autonomous technologies.
Research Focus
Key Achievements
Top Papers
- 1
- 2AdvRain: Adversarial Raindrops to Attack Camera-Based Smart Vision Systems19 citations · 2023
- 3SAAM: Stealthy Adversarial Attack on Monocular Depth Estimation12 citations · 2024
- 4
- 5
- 6