TUQCP combines PGD-based adversarial training, a learning-based uncertainty head, and conformal prediction to improve the accuracy and uncertainty estimates of collaborative object detection models under white-box attacks.
Adversarial objectness gradient at- tacks in real-time object detection systems
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Uncertainty Quantification for Collaborative Object Detection Under Adversarial Attacks
TUQCP combines PGD-based adversarial training, a learning-based uncertainty head, and conformal prediction to improve the accuracy and uncertainty estimates of collaborative object detection models under white-box attacks.