Alert button

Probing Contextual Diversity for Dense Out-of-Distribution Detection

Add code
Alert button
Aug 30, 2022
Figure 1 for Probing Contextual Diversity for Dense Out-of-Distribution Detection
Figure 2 for Probing Contextual Diversity for Dense Out-of-Distribution Detection
Figure 3 for Probing Contextual Diversity for Dense Out-of-Distribution Detection
Figure 4 for Probing Contextual Diversity for Dense Out-of-Distribution Detection

Share this with someone who'll enjoy it:

Detection of out-of-distribution (OoD) samples in the context of image classification has recently become an area of interest and active study, along with the topic of uncertainty estimation, to which it is closely related. In this paper we explore the task of OoD segmentation, which has been studied less than its classification counterpart and presents additional challenges. Segmentation is a dense prediction task for which the model's outcome for each pixel depends on its surroundings. The receptive field and the reliance on context play a role for distinguishing different classes and, correspondingly, for spotting OoD entities. We introduce MOoSe, an efficient strategy to leverage the various levels of context represented within semantic segmentation models and show that even a simple aggregation of multi-scale representations has consistently positive effects on OoD detection and uncertainty estimation.

* Safe Artificial Intelligence for Automated Driving Workshop, ECCV 2022  

Share this with someone who'll enjoy it: