TY - GEN
T1 - Enhancing Resilience
T2 - 2024 IEEE Space, Aerospace and Defence Conference, SPACE 2024
AU - Agarwal, Shubham
AU - Birman, Raz
AU - Hadar, Ofer
N1 - Publisher Copyright:
© 2024 IEEE.
PY - 2024/1/1
Y1 - 2024/1/1
N2 - Vision systems, like other deep learning-based systems, encounter limitations due to training data and struggle to handle adversarial conditions such as varying lighting and weather conditions. In this paper, we propose a knowledge distillation framework aimed at bolstering the resilience of computer vision systems under adversarial conditions. Specifically, we focus on object detection task in adverse weather conditions and demonstrate that our system either exceeds or matches the state-of-the-art accuracy levels. Our system achieves a 2% higher mean average precision (mAP@50) in hazy conditions, and 9% higher mean average precision (mAP@50) in low-light conditions, compared to the nearest state-of-the-art frameworks.
AB - Vision systems, like other deep learning-based systems, encounter limitations due to training data and struggle to handle adversarial conditions such as varying lighting and weather conditions. In this paper, we propose a knowledge distillation framework aimed at bolstering the resilience of computer vision systems under adversarial conditions. Specifically, we focus on object detection task in adverse weather conditions and demonstrate that our system either exceeds or matches the state-of-the-art accuracy levels. Our system achieves a 2% higher mean average precision (mAP@50) in hazy conditions, and 9% higher mean average precision (mAP@50) in low-light conditions, compared to the nearest state-of-the-art frameworks.
KW - adverse conditions
KW - computer vision
KW - knowledge distillation
KW - object detection
KW - redundancy reduction
UR - http://www.scopus.com/inward/record.url?scp=85205246981&partnerID=8YFLogxK
U2 - 10.1109/SPACE63117.2024.10668393
DO - 10.1109/SPACE63117.2024.10668393
M3 - Conference contribution
AN - SCOPUS:85205246981
T3 - 2024 IEEE Space, Aerospace and Defence Conference, SPACE 2024
SP - 580
EP - 583
BT - 2024 IEEE Space, Aerospace and Defence Conference, SPACE 2024
PB - Institute of Electrical and Electronics Engineers
Y2 - 22 July 2024 through 23 July 2024
ER -