Vos: Learning what you don't know by virtual outlier synthesis

X Du, Z Wang, M Cai, Y Li - ar** representations for detecting out-of-distribution objects
X Du, G Gozum, Y Ming, Y Li - Advances in Neural …, 2022 - proceedings.neurips.cc
Detecting out-of-distribution (OOD) objects is indispensable for safely deploying object
detectors in the wild. Although distance-based OOD detection methods have demonstrated …

Quantification of uncertainty and its applications to complex domain for autonomous vehicles perception system

K Wang, Y Wang, B Liu, J Chen - IEEE Transactions on …, 2023 - ieeexplore.ieee.org
Over the last decades, deep neural networks (DNNs) have penetrated all fields of science
and the real world. As a result of the lack of quantifiable data and model uncertainty, deep …

Normalizing flow based feature synthesis for outlier-aware object detection

N Kumar, S Šegvić, A Eslami… - Proceedings of the …, 2023 - openaccess.thecvf.com
Real-world deployment of reliable object detectors is crucial for applications such as
autonomous driving. However, general-purpose object detectors like Faster R-CNN are …

Can OOD Object Detectors Learn from Foundation Models?

J Liu, X Wen, S Zhao, Y Chen, X Qi - European Conference on Computer …, 2024 - Springer
Abstract Out-of-distribution (OOD) object detection is a challenging task due to the absence
of open-set OOD data. Inspired by recent advancements in text-to-image generative models …

Driver drowsiness detection and alert system

R Kannan, P Jahnavi, M Megha - 2023 IEEE International …, 2023 - ieeexplore.ieee.org
Driver fatigue is one of the leading causes of accidents worldwide. One of the most reliable
methods of measuring driver fatigue is to detect the driver's drowsiness. Drowsiness and …

Conformal prediction for trustworthy detection of railway signals

L Andéol, T Fel, F de Grancey, L Mossina - AI and Ethics, 2024 - Springer
We present an application of conformal prediction, a form of uncertainty quantification with
guarantees, to the detection of railway signals. State-of-the-art architectures are tested and …

Identifying out-of-distribution samples in real-time for safety-critical 2D object detection with margin entropy loss

Y Blei, N Jourdan, N Gählert - arxiv preprint arxiv:2209.00364, 2022 - arxiv.org
Convolutional Neural Networks (CNNs) are nowadays often employed in vision-based
perception stacks for safetycritical applications such as autonomous driving or Unmanned …

Confident Object Detection via Conformal Prediction and Conformal Risk Control: an Application to Railway Signaling

L Andéol, T Fel, F De Grancey… - Conformal and …, 2023 - proceedings.mlr.press
Deploying deep learning models in real-world certi ed systems requires the ability to provide
con dence estimates that accurately reflect their uncertainty. In this paper, we demonstrate …