Timezone: »

Instance-Aware Observer Network for Out-of-Distribution Object Segmentation
Victor Besnier · Andrei Bursuc · Alexandre Briot · David Picard

Recent works on predictive uncertainty estimation have shown promising results on Out-Of-Distribution (OOD) detection for semantic segmentation. However, these methods struggle to precisely locate the point of interest in the image, i.e, the anomaly. This limitation is due to the difficulty of fine-grained prediction at the pixel level. To address this issue, we build upon the recent ObsNet approach by providing object instance knowledge to the observer. We extend ObsNet by harnessing an instance-wise mask prediction. We use an additional, class agnostic, object detector to filter and aggregate observer predictions. Finally, we predict an unique anomaly score for each instance in the image. We show that our proposed method accurately disentangles in-distribution objects from OOD objects on three datasets.

Author Information

Victor Besnier (Valeo / Ecole des Ponts Paristech)

PhD student at Valeo and at the École des Ponts ParisTech (ENPC) in the Imagine lab. I am supervised by David Picard (ENPC) and Alexandre Briot (Valeo). I study the reliability of neural networks for the safety of autonomous vehicles. I am mainly focusing my research on Out-of-Distribution detection on image segmentation for safety-critical and real-time applications. I am also interested in generative approaches such as GAN, Normalizing Flows and Diffusion Models. Looking for job opportunities early 2023 :)

Andrei Bursuc (Valeo)
Alexandre Briot (Valeo)
David Picard (École des Ponts ParisTech)

More from the Same Authors