Information-Theoretic Segmentation by Inpainting Error Maximization

Autor:	Savarese, Pedro, Kim, Sunnie S. Y., Maire, Michael, Shakhnarovich, Greg, McAllester, David
Rok vydání:	2020
Předmět:	Computer Science - Computer Vision and Pattern Recognition
Druh dokumentu:	Working Paper
Popis:	We study image segmentation from an information-theoretic perspective, proposing a novel adversarial method that performs unsupervised segmentation by partitioning images into maximally independent sets. More specifically, we group image pixels into foreground and background, with the goal of minimizing predictability of one set from the other. An easily computed loss drives a greedy search process to maximize inpainting error over these partitions. Our method does not involve training deep networks, is computationally cheap, class-agnostic, and even applicable in isolation to a single unlabeled image. Experiments demonstrate that it achieves a new state-of-the-art in unsupervised segmentation quality, while being substantially faster and more general than competing approaches. Comment: Published as a conference paper at CVPR 2021
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2012.07287 Zobrazit plný text záznamu View this record from Arxiv